domenica 27 gennaio 2013

new 45 glomus tumors 01


Glomus Tumors

Anatomy
Glomus bodies are found in the jugular bulb and along the tympanic (Jacobson) and auricular (Arnold) branch of the tenth nerve in the middle ear or in other anatomic sites (Fig. 45.1). Depending on the location, glomus tumors (chemodectoma or paraganglioma) can be classified as tympanic (middle ear), jugulare, or carotid vagal, or designated as originating from other locations, such as the larynx, adventitia of thoracic aorta, abdominal aorta, or surface of the lungs (56,152) (Fig. 45.2). These tissues are responsive to changes in oxygen and carbon dioxide tensions and pH.


Figure 45.1. Anatomy of the region of the glomus jugulare.

Figure 45.2. Distribution of paragangliomas of the head and neck region.Laterality was not specified in three patients with carotid body paragangliomas. The diagram does not include one left carotid body paraganglioma that was found incidentally at autopsy and a left vagal body paraganglioma that presented in a patient who had two other paragangliomas.
Glomus tumors consist of large epithelioid (smooth muscle) cells with fine granular cytoplasm embedded in a rich capillary network and fibrous stroma with reticulin fibers, which derive from embryonic neural crest cells. Although histologically benign, they may extend along the lumen of the vein to regional lymph nodes, but rarely to distant sites.

Epidemiology
The mean age at diagnosis has been reported to be 44.7 years for carotid body tumors (196) and 52 years for glomus tympanicum (157). These tumors occur three or four times more frequently in women than in men, suggesting a possible estrogen influence (145,157,181). Glomus tumors may be familial; they also occur in multiple sites in 10% to 20% of patients (167,194).

Bilateral carotid glomus tumors were reported in six of 16 patients (38%) with a positive family history for these lesions but in only 17/206 patients (8%) without such a history (196). Multiple paragangliomas of the head and neck are rare (incidence of 10% of the total patients, but in familial cases it increases up to 35% to 50%) (167). In the head and neck region, the most common association is bilateral carotid body tumors or carotid body tumor associated with tympanic-jugular glomus.

Clinical Presentation
Glomus tumors may arise along the nerve roots. Glomus tumors of the middle ear may initially cause earache or discomfort. As they expand, eventually they produce pulsatile tinnitus, hearing loss, and, in later stages, cranial nerve paralysis resulting from invasion of the base of the skull in 10% to 15% of patients. Some patients endure ear symptoms for 3 to 5 years before seeking medical attention.

If the tumor invades the middle cranial fossa, symptoms may include temporoparietal headache, retro-orbital pain, proptosis, and paresis of cranial nerves V and VI. If the posterior fossa is involved, symptoms may include occipital headache, ataxia, and paresis of cranial nerves V to VII, IX, and XII; invasion of the jugular foramen causes paralysis of nerves IX to XI. Gaut et al. (95) described a case of a large intranasal glomus tumor that, at presentation, had eroded through the ethmoid roof to involve the floor of the anterior cranial fossa. The patient was treated with primary external-beam radiotherapy. To our knowledge, this is the first report of an invasive glomus tumor of the head and neck.

Chemodectoma of the carotid body usually presents as a painless, slowly growing mass in the upper neck. Occasionally the mass may be pulsatile and may have an associated thrill or bruit. As it enlarges, the mass may extend into the parapharyngeal space and be visible on examination of the oropharynx. Very rarely these tumors may be malignant (171).

Metastases occur in 2% to 5% of cases (145,220).

Diagnostic Work-Up
Diagnostic evaluation for glomus tumors of the ear and base of skull is outlined in Table 45.1. In the majority of glomus tympanicum tumors, physical examination demonstrates a red, vascular middle ear mass, although occasionally it may be bluish or white (the latter resembling a cholesteatoma) (157). Audiography may demonstrate conductive hearing loss in the ear involved by tumor as noted in 33/49 patients evaluated by Larson et al. (157); four of 33 patients with conductive deficits also exhibited tympanic pulsations. Examination of the neck may occasionally demonstrate a mass in the neck that may be pulsatile or have a bruit or regional lymph node metastases.


Table 45.1. Diagnostic Work-Up for Glomus Tumors of the Ear and Base of Skull, Hemangiopericytoma, Esthesioneuroblastoma, Extramedullary Plasmacytoma, and Sarcoma of the Head and Neck
Radiographic studies are invaluable in the diagnosis of these tumors. Plain mastoid radiographs never show the soft-tissue mass in the middle ear, although they frequently demonstrate clouding of the mastoid air cells, suggesting mastoiditis (157). High-resolution computed tomography (CT) with contrast has the highest degree of sensitivity and specificity to diagnose this tumor when located in the middle ear or jugular bulb; masses as small as 3 mm have been demonstrated in the middle ear. Tumor enhancement is similar to that of the temporalis muscle (157) (Fig. 45.3). In 46 patients with glomus tympanicum chemodectomas, there were no instances of local bony erosion; instead, the tumors engulfed the ossicular chain, bulged or protruded through the tympanic membrane, filled the middle ear, or extended into the eustachian tube orifice or aditus ad antrum. This pattern is in contrast to cholesteatomas, which typically destroy adjacent bony landmarks including the ossicles and progressively erode the petrous bones as they enlarge (157).



Figure 45.3.A: Late-phase arteriogram illustrating large glomus jugulare tumor with extension into the neck. B: CT scan with contrast enhancement showing intracranial component of lesion.
Magnification angiography is a sensitive and specific means of detecting glomus tympanicum tumors. This procedure should be performed after high-resolution thin-section CT scan (with contrast material), only when there is a question regarding the nature of the lesion or the location of the carotid canal. Findings include a hypervascular middle ear mass that first appears in the middle to late arterial phase, persists through the capillary phase, and quickly disappears in the venous phase without demonstrably early draining veins (162). Biopsy of an aberrant internal carotid artery can result in major neurologic sequelae or death.

Vogl et al. (275) reported on 40 patients with glomus tumors of the skull; diagnostic interpretations were correlated with histologic examination, digital subtraction angiography, CT, and clinical follow-up. Sixteen of 18 proven tumors were detected with spin-echo images alone. Although four high-flying jugular bulbs were misinterpreted as tumor because of similar signal intensity, combined evaluation allowed differentiation between tumor and sinusal blood flow in all cases.

Drape et al. (68) described magnetic resonance imaging (MRI) findings in 31 patients with a clinical suspicion of glomus tumor; gadoterate meglumine was injected in 19 patients. Twenty-seven of 28 pathologically confirmed glomus tumors were detected with MRI; a peripheral capsule was present in most tumors. The investigators were able to differentiate three subtypes of glomus tumors (vascular, solid, and myxoid) on the basis of relaxation times and enhancement characteristics.

Laird et al. (153) reported on 30 patients with neck masses; a bolus injection of 99mTc gluconate (20 mCi injected into the basilic vein) immediately followed by rapid injection of saline and scanning of the head and neck demonstrated glomus jugulare or carotid body tumors in seven patients, including two with clinically unsuspected tumors. The procedure was particularly useful in differentiating chemodectomas from other head and neck lesions such as thyroid tumors, parathyroid tumors, cystic hygromas, bronchogenic cysts, neural tumors, sarcomas, and lymph nodes.

Cytochemical techniques demonstrate increased levels of serotonin, epinephrine, and norepinephrine in normal glomus tissue of the carotid body. Histologic staining techniques, including chromaffin and argentaffin reactions, identify patients with hormonally active tumors. This is important because the glomus tumor may coexist with a pheochromocytoma, which requires special preoperative preparation of the patient.

Biopsy of glomus tumors may result in severe hemorrhage.

Staging
The prognosis of these tumors is closely related to the anatomic location and the volume of the lesion, which is reflected in the Glasscock-Jackson classification shown in Table 45.2. An alternative classification proposed by McCabe and Fletcher (176) is presented in Table 45.3.


Table 45.2. Glasscock-Jackson Classification of Glomus Tumors

Table 45.3. Modification of Mccabe and Fletcher Classification of Chemodectomas
General Management
Surgery
Surgery is generally selected for treatment of small tumors that can be completely excised. Glomus tympanicum tumors are particularly well managed with excision, via tympanotomy or mastoidectomy. Percutaneous embolization of a low-viscosity silicone polymer has been used, frequently as preoperative preparation of the tumor embolization of feeding vessels allows meticulous microsurgery with virtually complete hemostasis (194).

Surgical treatment of a glomus tumor arising in the jugular bulb, however, often consists of piece-by-piece removal accompanied by significant bleeding with damage to adjacent neurovascular structures and requires more complex surgical approaches involving the base of the skull. Intraoperative bleeding during surgical removal of head and neck paragangliomas may be a major problem in the management of these highly vascularized tumors. Preoperative embolization via a transarterial approach has proved beneficial but is often limited by vascular anatomy and unfavorable locations. Abud et al. (2) report experience with preoperative devascularization using direct puncture and an intralesional injection of cyanoacrylate (acrylic glue) under fluoroscopic guidance in nine patients with head and neck paragangliomas. Angiograms showed that complete devascularization was achieved in all cervical glomus tumors, whereas subtotal devascularization was achieved in jugular paragangliomas, because the injection of acrylic glue was limited by the potential risk of reflux into normal brain via feeders from the internal carotid or vertebral artery. The tumors were surgically removed.

The local tumor control rate with surgery alone is only about 60%, and there is significant morbidity, particularly cranial nerve injury and bleeding.

In a retrospective review of all skull-base surgery cases treated at Baylor University 175 jugulotympanic glomus tumors and nine malignant cases (5.1%) were identified (171). The 5-year survival rate was 72%.

Radiation Therapy
Irradiation is frequently used in the treatment of glomus tumors, particularly for those in the tympanicum and jugulare bulb locations. Tumors with destruction of the petrous bone, jugular fossa, or occipital bone or patients with jugular foramen syndrome are more reliably managed with irradiation (60,104,157,166,181,217,240). Some surgeons, such as Glasscock et al. (98) and Spector et al. (247), have questioned the effectiveness of radiation therapy in the treatment of chemodectomas because on histologic sections, obtained even many years after irradiation, it is possible to find chromophilic cells remaining in the tumor. However, there is also evidence of fibrosis and decreased vascularity (247). Suit and Gallager (256) demonstrated in a murine mammary carcinoma model that morphologically intact cells may have lost their reproductive ability after irradiation, which is the ultimate end point of cell killing. Furthermore, it is extremely unusual to observe clinical regrowth of a glomus tumor after irradiation, even if they do not regress completely.

Some reports describe successful combinations of surgery with either preoperative or postoperative irradiation (93,247), or preoperatively in an attempt to make an unresectable tumor operable, postoperatively when obvious tumor could not be resected.

Radiation Therapy Techniques
Radiation therapy techniques are determined by the location and extent of the tumor, which must be defined before treatment (53,185,249). Limited, usually bilateral, portals should be used for relatively localized glomus tumors, whether or not the treatment is combined with surgery (Fig. 45.4). Dickens et al. (63) used a three-field arrangement with a superior-inferior wedged and lateral open field, with a weighting of 1:1:0.33. Figure 45.5 shows superior-inferior 60-degree and 45-degree wedged filtered fields. Electrons (15 to 18 MeV) with a lateral portal or combined with cobalt-60 (60Co) or 4- to 6-MV photons (20% to 25% of total tumor dose) render a good dose distribution (Fig. 45.6). In patients in whom tumor has spread into the posterior fossa, it may be necessary to use parallel opposed portals with 6- to 18-MV photons. Treatment is given at the rate of 1.8 to 2 Gy tumor dose per day with five treatments per week for a total tumor dose of 45 to 55 Gy in 5 weeks. Three-dimention (3D) conformal radiotherapy (RT) or image-guided radiation therapy (IMRT) are highly desirable techniques to treat these tumors, with excellent dose distributions (see Fig. 45.6). Table 45.4 summarizes the doses of irradiation recommended by several investigators and the probability of tumor control (185,249,280).


Figure 45.4.A: Portal used for relatively localized glomus tumor. B: Simulation film of patient with glomus tumor. C: Isodose distribution of a mixed-beam unilateral portal for a glomus tympanicum lesion (80% 16-MeV electrons, 20% 4-MV photons).

Figure 45.5.Isodose distributions using superior-inferior pairs of 60-degree (A) and 45-degree (B) converging wedge filtered 60Co fields, demonstrating limited volume of irradiation.




Figure 45.6. Female 59-years-old with an unusual malignant left glomus jugulare, who had a metastatic left upper cervical lymph node.She was treated definitively with intensity-modulated radiation therapy (66 Gy in 2-Gy fractions). A: Cross, (B) coronal, and (C) sagittal sections showing dose distributions at primary site and left neck, sparing normal structures (D) dose-volume histogram:
Structure         Dose Range (Gy)       Mean Dose (Gy)
Planning target volume (including left neck)            38-77   70
Brain   0-59     2
Brainstem       6-35     12
Spinal cord     0-32     13

Table 45.4. Local Control with Radiation Therapy for Chemodectoma of the Temporal Bone (glomus tympanicum and Jugulare)
Leber et al. (158) reported on 13 patients with glomus tumors treated with radiosurgery because of recurrences after surgical removal in six patients. Histology was not available in seven patients, diagnosis was made from neuroradiological features only. Two patients had partial embolization before Gamma Knife (Elekta, Norcross, GA) treatment. Mean follow-up was 42 months (range, 14 to 72 months). Within the follow-up period there was no tumor progression and no clinical deterioration in any patient; 64% of the patients had an improvement of their symptoms, and in 36% the volume of the lesion decreased in size. There was no radiation-related morbidity.

Results of Therapy
The postirradiation change in tumor size is slow, with an increase in proliferative and perivascular fibrosis and minimal alterations in the chief epithelial cells (247). Histologic evaluation of tumor cell viability is not reliable (256). Despite the persistence of tumor both clinically and angiographically (166), amelioration of symptoms, absence of disease progression, and occasional return of cranial nerve function have been reported.

Seventeen patients were treated for glomus tympanicum tumors at Washington University (145). In five patients initial treatment consisted of irradiation alone, and all were tumorfree at last follow-up (4.5 years in one patient) or at death.Seven of eight patients irradiated for surgical recurrence were free of disease 4.5 to 19 years after irradiation. The remaining four patients were treated preoperatively or postoperatively; only one had recurrence and was salvaged surgically and tumorfree 10 years later. Of six patients with glomus jugulare lesions treated with irradiation, two with extensive lesions died of their disease, whereas the glomus tumor was controlled in four, including two patients with intracranial extension. Irradiation doses ranged from 46 to 52 Gy, with 86% to 100% tumor control with doses over 46 Gy and 50% (two of four) with doses below 46 Gy.

Of 19 patients treated with irradiation at the M.D. Anderson Cancer Center, five had only a biopsy without any surgical excision and 14 had partial excision (264). Ten patients had bony destruction; five of these had petrous pyramid and jugular foramen destruction, with accompanying multiple cranial nerve paralysis. Seventeen patients were treated with 60Co anteriorposterior or superior-inferior wedged filtered fields, and two patients received electrons and photons (3:1) via a single lateral field. Of 18 patients surviving a minimum of 5 years (13 surviving more than 10 years), all are alive and free of disease or have died of other causes.

Wang et al. (280) reported on 32 patients with tympanic chemodectomas; 13 treated with surgery alone, 15 with irradiation alone, and four with a combination of both modalities. The initial tumor control rate was 46% with surgery alone; ultimately 84% of patients were tumor-free after salvage with additional surgery. Although 78% survived 10 years, 31% developed complications. Of the patients treated with irradiation, 84% had initial local tumor control; 77% survived 10 years, and only 11% developed complications. The doses of irradiation used were slightly higher than those reported by others (mean 58.32 Gy). However, no improvement in tumor control was noted with higher doses. Complications occurred in two patients receiving 66 Gy.

In a compilation of several studies, Kim et al. (139) noted a 25% local failure rate in 83 patients treated with <40 Gy and 1.4% local failure in 142 patients receiving more than 40 Gy. Arthur (6) reported no recurrences in 24 patients treated with doses of 45 to 50 Gy; only one failure was observed in a patient receiving 30 Gy in 15 fractions in 21 days. If the tolerances of the brain and brainstem to irradiation are considered, doses of 50 Gy (1.8- to 2-Gy fractions) are considered optimal for treatment of these lesions.

Powell et al. (216) reported on 84 patients with chemodectoma of the head and neck, 46 of which were in the glomus jugulare and tympanicum, treated with irradiation alone (45 to 50 Gy in 25 fractions). Local control of the lesion was 73% at 5 years. Thirty patients were treated with surgery after irradiation with no recurrences (median follow-up of 9 years). Four patients, treated with surgery alone, developed recurrences by 7 years. Four carotid body and glomus vagal tumors treated with irradiation were locally controlled at 1, 2, 8, and 11 years, respectively. In 13 patients treated with surgery alone, the 15-year local control rate was 54%.

Mendenhall et al. (181) treated six chemodectomas of the carotid body and ganglion nodosum in four patients with doses of 40.8 to 48.5 Gy using 60Co, 8-MV x-rays, or a combination of 8- and 17-MV x-rays. Lesions have remained stable in four patients 2 to 4.5 years after irradiation. Treatment results in a few patients with this type of tumor are summarized in Table 45.5.


Table 45.5. Chemodectomas of the Carotid Body and Ganglion Nodosum
Hinerman et al. (119) reported on 71 patients with 80 chemodectomas of the temporal bone, carotid bone, or glomus vagal treated with radiation therapy alone in 71 patients or subtotal resection and radiation therapy (eight tumors). Fourteen patients had undergone a previous treatment (surgery 11, irradiation one, or both two). Fifty-three patients had temporal bone chemodectomas, 46 of which were classified as glomus jugulare and nine as glomus tympanicum. Pathologic confirmation of chemodectomas was obtained in 21 patients and the diagnosis was made on physical or radiographic findings in the remaining 32 patients. Fifty patients were treated with radiation therapy alone and five with subtotal resection followed by postoperative radiation therapy for gross residual tumor. Median dose was 45 Gy with daily fractions of 1.5 to 2 Gy delivered with 60Co, 6-MV, or 8-MV x-rays, or a combination of different beam energies. Twenty-eight patients were treated with ipsilateral wedge pair field arrangement, 18 with parallel-opposed fields; one patient was treated with stereotactic irradiation, and two were treated with three-dimensional conformal radiation therapy (3DCRT). Local control was obtained in 43 previously untreated lesions (93%) and in 11/12 (92%) previously chemodectomas. The results of treatment for temporal bone chemodectoma are summarized in Table 45.6.


Table 45.6. Temporal Bone Chemodectomas: Local Control After Radiation Therapy Alone or Radiation Therapy and Surgery
Eighteen patients had 25 chemodectomas of carotid body and/or glomus vagal; 15 tumors originated in the carotid body and 10 in the glomus vagal. Pathologic confirmation of chemodectoma was obtained in 10 patients, and diagnosis was based on physical and radiographic findings in the remaining eight. Twenty-two lesions were treated with radiation therapy alone, and two received postoperative radiation therapy after surgical resection for gross residual tumor with malignant changes and lymph node involvement. Patients with benign glomus tumors received 45 Gy in 25 fractions, in most instances, whereas patients with malignant carotid body tumors received 64.8 Gy to 70 Gy in 1.8 Gy fractions. Local tumor control was obtained in 14/15 carotid body and 10/10 glomus vagal (overall 96% tumor control) (274). The results of treatment for carotid body/glomus vagal are summarized in Table 45.7.


Table 45.7. Chemodectomas of Carotid Body/Glomus Vagale: Radiation Therapy Alone or Radiation Therapy After Surgery
Complications were rare in patients treated with chemodectoma of the head and neck.

Hemangiopericytoma

Hemangiopericytomas are rare soft-tissue neoplasms that account for 3% to 5% of all soft-tissue sarcomas and 1% of all vascular tumors. Some 15% to 30% of all hemangiopericytomas occur in the head and neck; of these, approximately 5% occur in the sinonasal area.

These tumors are believed to originate from the pericytes of Zimmerman extravascular cells morphologically resembling smooth muscle, found around the capillaries or from primitive mesenchymal cells. The function of the pericyte is uncertain but is believed to provide mechanical support for the capillaries having contractile function (254).

Epidemiology
Hemangiopericytomas is an unusual tumor; it represents approximately 1% of all vascular neoplasms; it occurs in both genders with equal frequency and is found primarily in adults. Only 45 cases of primary hemangiopericytomas of bone were described in the world literature in 1988.

In the head and neck, the most common sites are the nasal cavity and the paranasal sinuses, and usually, the orbital region, the parotid gland, and the neck (77,201,253).

Pathology
Hemangiopericytomas are composed of a proliferation of tightly packed pericytes around thin-walled endothelial-lined vascular channels ranging from capillary-sized vessels to large, gaping sinusoidal spaces (77). The tumor has a tendency to grow slowly and invade locally into adjacent structures. Although they are always well circumscribed and partially or completely surrounded by a pseudocapsule, benign tumors may be difficult to differentiate from malignant ones. However, prominent mitoses (greater than four per high-power field), foci of necrosis, and increased cellularity are suggestive of malignancy (77); the definitive sign is local recurrence or development of metastases. In general, tumors of the central nervous system, lower extremity, and mediastinum tend to be more malignant, with local recurrence occurring in up to 50% of cases (77). The final diagnosis of hemangiopericytomas is based on the histopathology and immunochemistry, and whether the tumor is benign or malignant is defined on the basis of the clinical history. Hemangiopericytomas located in the sinonasal area is generally benign.

Kowalski and Paulino (148) reviewed 12 cases of hemangiopericytomas. Proliferation index was assessed using an immunoperoxidase stain for MIB-1 (Ki-67). The mitotic index per 10 high power fields varied from 0 or 1 to 15. Proliferation indices using MIB-1 ranged from 2.6% to 52.5%. Clinical follow-up revealed three cases with recurrence all possessing proliferation indices of approximately 10%, indicating amore aggressive subset of hemangiopericytomas. Vuorinen et al. (278) found the proliferation index to be a poor predictor of prognosis.

Meningeal hemangiopericytomas almost always recur, despite seemingly complete removal due to infiltrative properties of hemangiopericytoma cells and not just higher proliferation potential. They often metastasize.

Clinical Presentation
Soft-tissue hemangiopericytoma is a firm, painless, slowly expanding mass that is often nodular and well localized. The skin overlying the mass does not have any discoloration or redness to indicate its vascular origin because the capillaries are emptied of the blood by compression of massive numbers of pericytes surrounding them (55,77).

In the head and neck, the tumor may constitute a polypoid, soft gray or red mass that grows slowly and may cause nasal obstruction. Epistaxis and nasal obstruction are common symptoms. The hemangiopericytomalike tumors of the nasal passages and paranasal sinuses differ slightly from those occurring elsewhere and probably represent a related but separate entity because they have little tendency toward recurrence or metastases regardless of the type of therapy.

Orbital hemangiopericytomas account for 3% of orbital malignancies and most frequently occur with painless proptosis (236). Hemangiopericytoma rarely originates in the lacrimal sac; it occurs in a younger age group than that of hemangiopericytoma of other locations. Charles et al. (42) reported on seven cases previously described and added one case.

Hemangiopericytoma may occur intracranially. When it arises in the brain, it is a solid mass attached to the meninges that grossly resembles a meningioma (128). These intracranial hemangiopericytomas carry a high risk of local failure (80%), as well as higher potential for dissemination. The mean time for local recurrence is 75 months (128).

The incidence of metastasis, which depends on the site of origin, can be 50% to 80%. Late metastases occurring 10 years after diagnosis are not uncommon.

On plain radiographs, hemangiopericytoma appears as a soft-tissue mass in the nasal cavity or other portions of the head and neck. A defect caused by pressure erosion of the surrounding bones may occur, and calcifications are rare. On arteriography, according to Yaghmai (293), hemangiopericytoma is the only vascular tumor that has some characteristic angiographic features that include radially arranged or spiderlike branching vessels around and inside the tumor and a long-standing, welldemarcated tumor stain. Intracranial tumors typically have arterial blood supply from both meningeal and cerebral connections, with one to three main feeders supplying many small corkscrewlike vessels (128). The most distinctive and constant feature of this tumor is its hypervascularity; this tissue characteristic also may be demonstrated with contrast-enhanced CT (201). Intracranially, the diffusely enhancing tumor may closely resemble a meningioma on CT. However, some CT signs may suggest hemangiopericytoma rather than meningioma: lack of calcification, scarce surrounding edema, and ringlike enhancement (198). Both CT and MRI scans are of special value in the delineation of the full extent of the tumor.

General Management
Complete surgical resection, if possible, combined with preoperative embolization of the tumor, is the treatment of choice. More extensive surgery is required in tumors that show features of malignancy. Many patients undergo surgical treatment after embolization of the feeding artery(ies).

For incompletely resected tumors, postoperative radiation therapy is used (184). The role of chemotherapy in this tumor is not well determined; a few reports have described partial tumor regression in some lesions treated with cytotoxic agents. Doxorubicin, alone or in combination-drug regimens, is the most effective agent for metastatic hemangiopericytoma, producing complete and partial remissions in 50% of cases (290). Other drugs prescribed when metastasis occurs are cyclophosphamide, dacarbazine, vincristine, and actinomycin-D (113).

Radiation Therapy Techniques
The role of radiation therapy alone in the management of hemangiopericytoma is controversial. The main role of irradiation is as an adjuvant after complete excision of the lesion or postoperatively for minimal residual disease (79,131,172,248). The tumor has been considered relatively radioresistant. Tumor doses of 60 to 65 Gy in 6 to 7 weeks are required to produce local tumor control in postoperative cases (128). Orbital hemangiopericytoma has been cured by surgery and postoperative irradiation to 65 Gy (236).

There appears to be a definite role for postoperative irradiation to the brain for primary hemangiopericytoma when radical surgery is performed because these tumors tend to recur after seemingly complete removal. Jha et al. (131) reported local tumor control in all patients treated with adjuvant externalbeam irradiation postoperatively. Radiation therapy also has been used as a salvage procedure after local recurrence following initial surgery and/or chemotherapy.

The fields of irradiation should be wide, to encompass the tumor bed with a margin of at least 5 cmto safely avoid marginal recurrence. Portal arrangement and beam selection are similar to those used in treatment of malignant brain tumors or softtissue sarcomas.

Results of Therapy
Billings et al. (26) reported on 10 patients with hemangiopericytoma of the head and neck; seven tumors arose from softtissue sites and three from the mucosa. All patients underwent wide excision of the primary lesion with a local recurrence rate of 40%. Three patients developed metastatic lung disease 0 to 8 years after initial diagnosis. Each patient who developed metastatic disease had abundant mitoses on pathological review compared with rare or absent mitoses in the lesions that took a more benign course.

Patrice et al. (208) reported on 18 primary hemangioblastoma tumors (16 had no prior surgical resection and two were subtotally resected lesions) and 20 lesions treated after surgical failure with stereotactic irradiation (radiosurgery). Minimum tumor doses ranged from 12 to 20 Gy (median 15.5 Gy). With a median follow-up of 24.5 months (range 6 to 77 months), the 2-year actuarial survival was 88%, and the 3-year freedom from progression was 86%. Four of 22 patients died. Thirty-one of 36 evaluable tumors (86%) were controlled locally. None of the 18 primary tumors treated with definitive stereotactic irradiation failed. Of the 18 recurrent tumors, 13 (72%) were controlled. The median tumor volume of lesions that failed to be controlled was 7.85 cm3 compared with 0.67 cm3 for controlled lesions (p = 0.0023). There were no significant permanent complications attributable to the stereotactic irradiation.

Spitz et al. (248) published a report on 36 patients (older than 16 years) with hemangiopericytoma treated at the University of Texas M.D. Anderson Cancer Center. The median follow-up was 57 months. Twenty-eight patients (78%) underwent complete and potentially curative resection. Of the nine patients (32%) who had local recurrences, four (44%) had epidural tumors and three (33%) had retroperitoneal tumors, but none had extremity tumors. Ten patients had recurrences at distant sites. Of the 13 patients who experienced any form of disease recurrence, four had recurrences after a diseasefree interval of more than 5 years. The 5-year actuarial survival rate for the entire group of 36 patients was 71%.

Carew et al. (36) reviewed the records of 12 patients with hemangiopericytomas of the head and neck. Five patients had lesions characterized as high or intermediate grade histologically, and seven had low-grade lesions. Nine patients were treated with curative intent; three presented either with pulmonary metastasis (two patients) or unresectable primary tumor (one patient) and were treated with radiation therapy and/or palliative doxorubicin-based chemotherapy. Patients treated with curative intent underwent a variety of surgical resections dictated by tumor location and size. Four patients received postoperative radiation therapy to a median dose of 60 Gy, for positive surgical margins (two patients), high-grade histology (one patient), or a recurrent lesion (one patient). The 5-year overall survival rate for patients treated surgically was 87.5%. A single mortality occurred in a patient with a recurrent high-grade lesion who failed at local, regional, and distant sites.

Payne et al. (209) described their experience in 12 patients with 15 intracranial hemangiopericytomas treated using gamma surgery. Clinical and radiographic follow-up of 3 to 56 months was available for 10 patients with 12 tumors. There was one tumor present at the time of initial gamma surgery in each patient. Two new tumors occurred in patients previously treated. Nine of the tumors decreased in volume and three remained stable. Four of the nine tumors that shrank later progressed at an average of 22 months after treatment. There were no complications and the quality of life following the procedure was maintained or improved in every case.

Chordomas

Anatomy
Chordomas are rare neoplasms of the axial skeleton that arise from the remnant of the primitive notochord (chorda dorsalis). About 50% arise in the sacrococcygeal area; 35% arise intracranially, where they typically involve the clivus, and the remaining 15% occur in the midline along the path of the notochord, primarily involving the cervical vertebrae (259).

Epidemiology
Chordomas are more common in patients in their 50s and 60s but can occur in all age groups. In children and young adults the prognosis and long-term survival appear to be better than in older patients (289). No risk factors have been identified. Male predominance is reported at a 2:1 to 3:1 ratio.

Natural History
Although slowly growing, chordomas are locally invasive, destroying bone and infiltrating soft tissues. Basisphenoidal chordomas tend to cause symptoms earlier and may be difficult to differentiate histologically from chondromas and chondrosarcomas and radiographically from craniopharyngiomas, pineal tumors, and hypophyseal and pontine gliomas. The lethality of these tumors rests on their critical location, aggressive local behavior, and extremely high local recurrence rate. The incidence of metastasis, which has been reported to be as high as 25%, is higher than previously believed and may be related to the long clinical history. The most common site of distant metastasis is the lungs, followed by liver and bone. Lymphatic spread is uncommon.

Pathology
Chordoma is a soft, lobulated tumor that may have areas of hemorrhage, cystic changes, or calcification. It is frequently encapsulated but may be nonencapsulated or pseudoencapsulated. Histologically, it is composed of cords or masses of large cells (physaliferous cells) with typical vacuoles and granules of glycogen in the cytoplasm and abundant intercellular mucoid material. Usually there are few mitotic cells. Heffelfinger et al. (114) postulated that a chondroid variant of chordoma may exist, being prevalent in the spheno-occipital area. Patients with this type of histologic variant have improved survival.

Aside from the previously mentioned histologic features, the prognostic factors that most influence the choice of treatment are location and local extent of tumor.

Clinical Presentation
Chordomas tend to originate from the clivus and chondrosarcomas from the temporal bone (141). Clinical symptoms vary with the location and extent of the tumor. In the head, extension may be intracranial or extracranial, into the sphenoid sinus, nasopharynx, clivus, and sellar and parasellar areas, with a resultant mass effect. In chordomas of the spheno-occipital region, the most common presenting symptom is headache. Other presentations include symptoms of pituitary insufficiency, nasal stuffiness, bitemporal hemianopsia, diplopia, and other cranial nerve deficits. Fuller and Bloom (88) reported on 13 patients with clivus chordoma, all of whom had multiple cranial nerve palsies. Facial pain was present in 11/13 patients.

Volpe et al. (277) reviewed the clinical features of 48 patients with chordoma and 49 patients with low-grade chondrosarcoma of the skull base. Twenty-five patients (52%) with chordoma and 24 patients (49%) with chondrosarcoma had ocular symptoms (diplopia or visual impairment) as the initial manifestation of the disease. Of the 59 patients (both groups) with diplopia, the diplopia was initially intermittent in 25 (42%). Headache and diplopia from abducens nerve palsy occurred in 22 patients (46%) with chordoma and 23 (47%) with chondrosarcoma.

Diagnostic Work-Up
The diagnostic work-up varies with the primary location of disease. Most patients have significant bony destruction, and some may have calcifications in the tumor; hence, plain films and, specifically, CT scans or MRI are very useful (67) (Table 45.8). In most cases, the soft-tissue component is much more extensive than initially appreciated, and a CT scan with contrast enhancement is required (Fig. 45.7A). CT and MRI are equivalent for demonstration of the presence and site of these tumors. MRI is inferior to CT in its ability to demonstrate bony destruction and intratumoral calcification (Fig. 45.7B) (197,258). MRI is superior to CT regarding the delineation of the exact extent of the tumor, which allows for better treatment planning (67). Because of availability and lower cost, CT appears to be the technique of choice for routine follow-up of previously treated patients (197).


Table 45.8. Diagnostic Work-Up for Chordoma



Figure 45.7.A: Contrast material-enhanced axial computed tomography scan demonstrates a large chordoma with extension into the posterior fossa and left parasellar region. B: Computed tomography scan photographed at bone windows shows the bony destruction and intratumoral calcifications. C: Treatment planning field arrangement for illustrated clivus chordoma using standard irradiation techniques with wedges on lateral points.
Reliable signs of chordoma of the skull base are posterior extension to the pontine cistern; a lobulated, “honeycomb” appearance after gadolinium; the swollen appearance of the bone in the early stages; bone erosion on CT; and frequent extension to critical structures such as the circle of Willis, cavernous sinuses, and brainstem (67).

General Management
Because of their surgical inaccessibility and relative resistance to radiation therapy, clivus chordomas represent a formidable therapeutic challenge. The general management of the patient is dictated by the anatomic location of the tumor and the direction and extent of spread. A surgical approach is recommended (when feasible), but complete surgical extirpation alone is unusual (230). Regression of preoperative symptoms without additional postoperative morbidity could be achieved by radical transoral tumor extirpation documented by MRI (234). Intracranial spread usually requires steroid coverage and therapy directed to correction of neurologic deficits that may be present. Because of the high incidence of local recurrence, combined surgical excision and irradiation is frequently used. No effective chemotherapeutic agent or combination of drugs has been identified.

Radiation Therapy Techniques
Irradiation techniques vary considerably, depending on the location of the tumor along the craniospinal axis. Basisphenoidal tumors usually are treated by a combination of parallel opposed lateral fields, anterior wedges, and photon and electron beam combinations, depending on the extent of the neoplasm. Precision radiation therapy planning, using CT and MRI, is required because high doses of external-beam radiation therapy are needed. Three-dimensional CRT or IMRT provide optimal dose distributions (Fig. 45.8).




Figure 45.8. Chordoma of clivus in 81-year-old man treated with 70 Gy in 2-Gy fractions.Example of IMRT plan: A: Cross-section in upper portion of planning target volume (PTV), demonstrating coverage of target volume with sparing of ocular structures. B: Sagittal plane dose distribution with excellent coverage of PTV. C: Dose-volume histogram:
Structure         Dose Range (cGy)      Mean Dose (Gy)
PTV (including left neck)      60-75   70
Optic nerves/chasm    25-50   41
Ocular globe   3-30     12
The tumor usually surrounds the spinal cord and infiltrates vertebral bones. A combined technique using protons or electrons to boost the initial photon fields is generally applied. In the treatment of chordomas surrounding the spinal cord, IMRT can provide high-dose homogeneity and planning target volume (PTV) coverage. Frequent digital portal image-based setup control reduces random positioning errors for head and neck cancer patients immobilized with conventional thermoplastic masks. Gabriele et al. (90) treated a patient with incomplete resection of a vertebral chordoma surrounding C2-3 with a total dose of 58 Gy (International Commission of Radiation Units and Measurements point) in 2-Gy daily fractions. Beam arrangement consisted of seven 6 MV nonopposed coplanar IMRT fields using 120-leaf collimator in sliding window mode. To verify the daily setup, portal images at 0 degrees and 90 degrees were compared with the simulation images before treatment delivery (manual matching) and after treatment delivery (automatic anatomy matching). The mean dose to the PTV was 57.6 ± 2.1 Gy covering 95% of the PTV with the 95% isodose. The minimum dose to the PTV (D99) was 53.6 Gy in the overlapping area between the PTV and the spinal cord planning organ at risk volume (PRV). The maximum dose to the spinal cord was 42.2 Gy and to the spinal cord PRV (8 mm margin) 53.7 Gy. The mean dose to the parotid glands were 37.4 Gy (homolateral gland) and 19.5 Gy (contralateral gland). Average deviation in setup was -1.1 ± 2.5 mm (anterior-posterior), 2.4 ± 1.3 mm (laterolateral), 0.7 ± 0.9 mm (craniocaudal) and –0.43 ± 1 degree (rotation).

Because of the slow proliferative nature of chordomas, high linear energy transfer may prove useful in their management, as it will be discussed later. Brachytherapy can be used for recurrent tumors of the base of skull or adjacent to the spine when a more aggressive surgical exposure is offered. Three of five chordomas were rendered stable when treated with iodine-94 (94I) implants by Gutin et al. (105), performed with CT stereotactic technique. Kumar et al. (150) reported use of 94I intraoperative interstitial implantation in two patients with recurrent chordomas. Disease was effectively controlled in both.

Results of Therapy
Photons
Although survival in some patients with chordoma may be long term, the salient feature of this unusual neoplasm is local recurrence with eventual death. The course may be indolent, with multiple treatments for recurrences, but the overall 5-year diseasefree survival rate is <10% to 20%. At M.D. Anderson Cancer Center, of 19 patients treated definitively, three were alive and free of disease with relatively short follow-up of 3, 6, and 7.5 years, respectively. Fuller and Bloom (88), in 25 patients treated with external-beam irradiation, found 96% stabilization or reduction of pain; the overall actuarial survival rates were 44% and 17% at 5 and 10 years, respectively

Catton et al. (37) analyzed the long-term results of treatment for patients with chordoma of the sacrum, base of skull, and mobile spine treated predominantly with postoperative photon irradiation. In 20 base of skull chordomas, most of them irradiated with conventionally fractionated radiation to a median dose of 50 Gy in 25 fractions for 5 weeks (range 25 Gy to 50 Gy), median survival was 62 months (range 4 to 240 months) from diagnosis with no difference between clival and nonclival presentations. There was no survival advantage to patients receiving radiation doses >50 Gy (median 60 Gy) compared with lower doses <50 Gy (median 40 Gy). Hyperfractionation regimens did not influence the degree or duration of symptomatic response or progressionfree survival. Median survival after retreatment was 18 months.

Forsyth et al. (86) reported on 51 patients with intracranial chordomas (19 classified as chondroid) treated surgically (biopsy in 11 patients and subtotal removal or greater in 40); 39 patients received postoperative irradiation. At the time of the analysis, 17 patients were alive. The 5- and 10-year survival rates were 51% and 35%, respectively; 5-year survival was 36% for biopsy patients and 55% for those who had resection. Patients who underwent postoperative irradiation tended to have longer diseasefree survival times.

Gay et al. (96) analyzed the outcome of 46 patients with cranial base chordomas and 14 with chondrosarcomas treated with extensive surgical resection: 50% of them had been treated previously; 20% received postoperative irradiation. Nine patients with chordomas and two with chondrosarcomas died during the postoperative follow-up period. The 5-year recurrencefree survival rate for all patients was 76%. Chondrosarcomas had a better prognosis than chordomas (5-year recurrencefree survival rates of 90% and 65%, respectively) (p = 0.09). Patients who had undergone previous surgery had a greater risk of recurrence than did those who had not undergone previous surgery (5-year recurrencefree survival rates of 64% and 93%, respectively) (p <0.05). Those with total or near-total resection had a better 5-year recurrencefree survival rate (84%) than did patients with partial or subtotal resection (64%) (p <0.05). Postoperative leakage of cerebrospinal fluid was the most frequent complication (30% of patients) and was found to increase the risk of permanent disability. Patients who had undergone previous irradiation had a greater risk of death in the postoperative period (within 3 months of operation) and during follow-up.

Tai et al. (259) reviewed the results of irradiation combined with surgery, irradiation alone, and surgery alone in 159 patients reported in the literature. An analysis of the optimal biologically equivalent dose was performed using the linearquadratic formula on 47 patients. With conventional photon irradiation no dose-response relationship was shown. Survival improved in patients undergoing surgery followed by irradiation.

Chetty et al. (45) reported on 18 chordomas, 61% of them occurred in the sphenoid region. Follow-up for 12 patients ranged from 3 to 170 months. Various combinations of surgery and radiation therapy were used. Mean survival was 73.4 months, with a survival rate of 50% (six of 12 patients).

Keisch et al. (138) reported on 21 patients with chordoma treated at our medical center: five had clival tumors, two had nasopharyngeal tumors, and one had a lumbar spine tumor. Nine patients were treated with surgery alone, eight had subtotal resection and postoperative irradiation, and four received irradiation alone after biopsy. The 5- and 10-year actuarial survival was significantly better in patients treated with surgery alone or surgery and irradiation than in those treated with radiation therapy alone (52%, 32%, and 0%, respectively) (p = 0.02). Diseasefree survival of patients with base of skull tumors was not significantly different among the treatment groups.

Debus et al. (59) reported on 45 patients treated for chordoma or chondrosarcoma with postoperative fractionated 3D stereotactic radiation therapy. Median dose at isocenter was 66.6 Gy for chordomas and 64.9 Gy for chondrosarcomas. All chondrosarcomas achieved and maintained local tumor and recurrencefree status at 5-years follow-up. Local control rate of chordomas at 5 years was 50% and survival was 82%. Clinically significant late toxicity developed in only one patient.

Kondziolka et al. (144) assessed the use of radiosurgery in four patients with chordoma and two with chondrosarcoma (in five patients as adjuvant therapy for residual or recurrent tumors after surgical debulking; in one patient with a chordoma as primary treatment). No patient received fractionated externalbeam irradiation. All tumors were <30 mm in diameter and were treated with 20 Gy to the tumor margin. During follow-up (mean 22 months; range 8 to 36 months), they found no progression of the treated tumor in any patient. Neurologic deficits before treatment improved in three patients; the other three patients remained in stable neurologic condition. Serial follow-up imaging studies demonstrated reduction in tumor size in two patients, and four patients had no tumor growth. One patient showed tumor progression outside the radiosurgical treatment volume.

Protons
The best results in the treatment of chordomas have been obtained with radical surgical procedures followed by high-dose proton irradiation. Berson et al. (22) described 45 patients with chordomas or chondrosarcomas at the base of the skull or cervical spine who were treated by subtotal resection and postoperative irradiation. Twenty-three patients were treated definitively by charged particles, 13 patients with photons and particles, and nine were treated for recurrent disease. Doses ranged from 36 to 80 Gy equivalent. There appeared to be significant benefit for patients with smaller tumor volumes (80% vs. 33% actuarial survival rate at 5 years). Patients treated for primary disease had a 78% actuarial local control rate at 2 years, whereas the rate for patients with recurrent disease was 33% (21).

Tatsuzaki and Urie (261) described the use of proton beam therapy at high doses for chordomas and chondrosarcomas of the base of the skull and cervical spine. Treatment delivered 74 cobalt gray equivalent (CGE) to the tumor while maintaining the central brainstem and central spinal cord at 48 CGE or less; and the surface of the brainstem, spinal cord, and optic structures at 60 CGE or less. Proton beam plans and 10-MV x-ray beam plans were developed with these assumptions and dose constraints. In all cases the proton beam plans delivered more dose to a larger percentage of the tumor volume, and the estimated tumor control probability was higher than with the x-ray plans. However, without precise positioning both the proton plans and the x-ray plans deteriorated, with a 12% to 25% decrease in estimated tumor control probability.

O'Connell et al. (193) reported on 62 patients with base of skull chordomas treated with proton beam irradiation (65 to 73.5 Gy equivalent); 29 patients (19 women and 10 men) experienced local failure, and 14 women (48%) and seven men (21%) died of disease. On histologic analysis, presence of more than 10% necrosis, prominent nucleoli, and tumor larger than 70 mm were significant predictors of short-term disease-specific survival. Chondroid chordoma and conventional chordomas had equivalent outcome.

Proton beam boosts have been recommended. Rich et al. (223) reported results in 48 patients with chordomas: 14 patients were treated with surgery and 17 with combination of partial surgical resection and irradiation, or irradiation alone after biopsy (15 patients) (Table 45.9). Various techniques were used to deliver doses of 45 to 80.4 Gy with photons alone or combined with 160-MeV protons, usually 2 Gy daily.


Table 45.9. Patient Status Correlated with Treatment and Radiation Dose Level in Chordoma
Fagundes et al. (80) updated the Massachusetts General Hospital experience with 204 patients treated for chordoma of the base of the skull or cervical spine. Sixty-three patients (31%) had treatment failures, which were local in 60 patients (29%) and the only site of failure in 49 patients. Two patients had regional lymph node relapse, and three developed surgical pathway recurrence. Thirteen patients relapsed in distant sites (especially lungs and bones). The 5-year actuarial survival rate after any relapse was 7%. There was no significant difference in survival for patients who had a local or distant failure. Two patients (1.4%) with local tumor control developed distant metastases in contrast with 10/60 patients (16%) who failed locally and distantly.

Terahara et al. (262) reported on 132 patients with skull base chordoma treated with combined photon and proton irradiation; in 115 patients dose-volume data and follow-up were available. The prescribed doses ranged from 66.6 CGE to 79.2 CGE (median of 68.9 CGE). The dose to the optic structures (optic nerves and chiasm), the brainstem surface, and the brainstem center was limited to 60, 64, and 53 CGE, respectively. Local failure developed in 42/115 patients, with the actuarial local tumor control rates at 5 and 10 years being 59% and 44% respectively. In a Cox multivariate analysis, the models equivalent uniform dose (EUD) suggest that the probability of recurrence of skull-base chordomas depends on gender, target volume, and target dose inhomogeneity; EUD was shown to be a useful parameter to evaluate dose distribution for the target volume.

Hug et al. (123) analyzed treatment efficacy of fractionated proton radiation therapy administered for skull base 33 chordomas and 25 chondrosarcomas. Following various surgical procedures, residual tumor was present in 91% of patients; 59% demonstrated brainstem involvement. Target doses ranged from 64.8 to 79.2 (mean 70.7) CGE. The range of follow-up was 7 to 75 months (mean, 33 months). In 10 patients (17%) the treatment failed locally, resulting in local control rates of 92% (23/25 patients) for chondrosarcomas and 76% (25/33 patients) for chordomas. All tumors with volumes of 25 mL or less remained locally controlled compared with 56% of tumors larger than 25 mL (p = 0.02). Of patients without brainstem involvement 94% did not experience recurrence; whereas with brainstem involvement (and dose reduction because of brainstem tolerance constraints) the tumor control rate was 53% (p = 0.04). Actuarial 5-year survival rates were 100% for patients with chondrosarcoma and 79% for patients with chordoma. Grade 3 and 4 late toxicities were observed in four patients (7%) and were symptomatic in three (5%).

Benk et al. (18) described results in 18 children 4 to 18 years of age with base of skull or cervical spine chordomas who received fractionated high-dose postoperative irradiation using mixed-photon and 160-MeV proton beams. Median tumor dose was 69 CGE with a 1.8-CGE daily fraction. With a median followup of 72 months, the 5-year survival was 68%, and the 5-year diseasefree survival rate was 63%. Patients with cervical spine chordomas had a worse survival rate than did those with base of skull lesions (p = 0.008). The incidence of treatment-related morbidity was acceptable: two cases of growth hormone deficit corrected by hormone replacement, one temporal lobe necrosis, and one fibrosis of the temporalis muscle, improved by surgery.

A report on proton therapy for base of skull chordoma was published by the Royal College of Radiologists (228). They concluded that outcome after proton irradiation is superior to that reported for conventional photon irradiation. Radiation therapy schedules involving a mixed schedule of protons and photons have achieved an approximately 60% local tumor control rate at 5 years.

Sequelae of Treatment
In patients treated with high irradiation doses, as well as with charged particles, there is an increasing probability of sequelae, including brain damage, spinal cord injury, bone or soft-tissue necrosis, and xerostomia. In a report by Berson et al. (22), three patients experienced unilateral visual loss, and four patients had radiation injury to the brainstem.

Santoni et al. (231) reported on the temporal lobe damage rate in 96 patients (75 primary and 21 recurrent tumors) treated with postoperative high-dose proton and photon irradiation for chordomas and chondrosarcomas of the base of the skull. All the patients were randomized to receive 66.6 or 72 CGE with conventional fractionation (1.8 CGE per day, five fractions a week) using opposed lateral fields for the photon component and a noncoplanar isocentric technique for the proton component. Of the 96 patients, 10 developed temporal lobe damage, (lateral in two and unilateral in eight). The cumulative temporal lobe damage incidence at 2 and 5 years was 7.6% and 13.2%, respectively. CT and MRI scans were evaluated for white matter changes; the MRI areas suggestive of temporal lobe damage in 10 patients were always separate from the tumor bed.

In patients receiving high-dose proton therapy for clivus tumors, Slater et al. (242) observed a 26% incidence of endocrine abnormalities at 3 years and 37% at 5 years, with hypothyroidism being the most frequent sequela. The dose to the pituitary in patients with abnormalities ranged from 63.1 to 67.7 Gy equivalent.

Lethal Midline Granuloma

Natural History and Pathology
Lethal midline granuloma (LMG) or midline malignant polymorphic reticulosis is a clinical entity characterized by progressive, unrelenting ulceration and necrosis of the midline facial tissues. LMG is associated with Epstein-Barr virus, which has at least two subtypes with different biologic properties that can be identified by their genomic configuration. The occurrence of the rare subtype 2 in LMG may relate to a covert immune defect (28). Considerable controversy exists regarding various disorders characterized by a necrotizing and granulomatous inflammation of the tissues of the upper respiratory tract and oral cavity. It is now clear that if infections and other known agents such as cocaine use, sarcoidosis, environmental toxins, and various neoplasms can be excluded, three clinicopathologic entities remain: Wegener's granulomatosis, LMG, and polymorphic reticulosis (PMR) (12). A review of the literature suggests that cases described as idiopathic midline destructive disease and PMR are a large evolutionary spectrum from almost benign to fatal malignant lymphoma (21).

Wegener's granulomatosis is an epithelioid necrotizing granulomatosis with vasculitis of small vessels. Systemic involvement of the kidneys and lungs is common.

PMR is an unusual disorder with distinctive clinical and pathologic features (177). Histologically, PMR is characterized by an atypical mixed lymphoid infiltration of the submucosa with extensive areas of necrosis, sometimes extending to bone or cartilage. The lesion consists of variable zones of small lymphocytes with scattered immunoblastic forms, abundant plasma cells with occasional eosinophilia, and histiocytosis (243). PMR has been considered a lymphoproliferative disorder; most, if not all, cases are peripheral T-cell lymphomas (163,283). Several authorities believe that PMR and systemic lymphomatoid granulomatosis are the same disease with the latter predominantly involving the lungs (94,163).

Idiopathic LMG describes a localized disorder not characterized by visceral lesions but by destruction of the midfacial area, which, if left untreated, is uniformly fatal. The histopathologic findings are nonspecific, with a relatively nondescript inflammatory reaction with acute and chronic inflammation and necrosis. Despite specific clinicopathologic features, the distinction between LMG and PMR is often difficult; although controversial, they may represent two phases of the same disease, with LMG remaining histologically benign or evolving into PMR. LMG occurs more frequently in men (94). Ages range from 21 to 64 years; almost half of the patients are in their 50s at presentation. Most patients have involvement of the nasal cavity (including destruction of the septum) and the paranasal sinuses (particularly maxillary antrum). The primary lesion may extend into the orbits, the oral cavity (palate, gingiva), and even the pharynx.

Characteristics of the three different diseases are outlined in Table 45.10.


Table 45.10. Differential Features of Three Clinicopathologic Entities
Clinical Features and Diagnostic Work-Up
Clinical manifestations include progressive nasal discharge, obstruction, foul odor emanating from the nose, and, in later stages, pain in the nasal cavity, paranasal areas, and even in the orbits.

Examination discloses ulceration and necrosis in the nasal cavity, perforation or destruction of nasal septum and turbinates, and even ulceration of the nose. Edema of the face and eyelids may be noted, and the bridge of the nose may be sunken. Radiographic studies initially show soft-tissue swelling, mucosal thickening, and findings consistent with chronic sinusitis.

CT is invaluable in demonstrating the full extent of the tumor, including bone or cartilage destruction. In 13 patients presenting with LMG, CT proved essential for determining the extent of the disease, guiding biopsy, and planning radiation therapy (173). MRI was also helpful for the latter because it could distinguish fluid retained within the paranasal sinuses from solid masses and tumor from granulation tissue; it was of little value for detecting bone lysis. Eight patients proved to have T-cell lymphoma, two had Crohn's disease, in one the lesion was factitious, and two had granulomas without diagnostic histologic features.

General Management and Radiation Therapy Techniques
When treatment of these patients is planned, it is extremely important to exclude the diagnosis of Wegener's granulomatosis, a benign process that is commonly treated with antimicrobial agents, steroids, and systemic chemotherapy (94,190). Bona fide LMG does not respond to steroids; the treatment of choice is radiation therapy (64,89,237).

Target volume should encompass all areas of involvement, including adjacent areas at risk (i.e., for a lesion of the maxillary antrum it will include the antrum as well as all of the paranasal sinuses) with a 2- to 3-cm margin (106). Because marginal failures are a significant problem, wide margins are necessary for treatment of these patients (243).

Irradiation techniques are similar to those described for tumors of the paranasal sinuses, nasal cavity, or nasopharynx. Several investigators have described complete responses with doses of 30 to 50 Gy; most patients are treated with 35 to 45 Gy in 3 to 4.5 weeks (64,83,84,237). We recommend 45 to 50 Gy in 4.5 to 5.5 weeks in 1.8- to 2-Gy daily fractions.

Results of Therapy
Because of the rarity of this tumor, experience is limited. Fauci et al. (84) reported on 10 patients with extensive midline granuloma treated with irradiation. Three received 10 Gy, and all failed within 2 years (retreated with 40 to 46 Gy). The remaining seven patients received 40 to 50 Gy. Local control of disease was 77%; two patients had local recurrences, one outside the initially irradiated volume.

The Mayo Clinic reports the most extensive experience in treatment of PMR or LMG with irradiation doses of 40 to 42 Gy (177). Of 20 patients irradiated for localized upper airway PMR, 13 were alive and well for an average of 9.5 years; two were alive and well with <1 year of follow-up; four were dead of other disease, and one was lost to follow-up.

In a study of 34 patients with PMR treated with primary radiation therapy except for one patient, Smalley et al. (243) found that a minimum dose of 42 Gy or a time-dose factor of 70 was necessary to achieve long-term local control. The most frequent failure site was within the original irradiation field. They believe that this problem should become much less significant with implementation of proper time-dose-fractionation schemes. Systemic failure occurred in 25% of their patients initially presenting with limited disease. The salvage of this subset of patients requires effective systemic chemotherapy. Also, Itami et al. (127) evaluated nine patients with locally confined nasal non-Hodgkin's lymphoma (NHL) treated with radiation therapy (all NHLs had T-lineage). Additionally, unique histological pictures of polymorphism, angiodestruction, and necrosis were seen in most cases, findings that are the histological features of PMR, which is the main cause of LMG. Although the disorder was considered to be locally limited at presentation, only three of the nine patients with nasal NHL could be induced into long-term remission with involved field radiotherapy (40 to 60 Gy) and distant extranodal spread was the primary cause of failure. Multimodality treatment using intensive chemotherapy and radiation therapy might improve the prognosis of these patients.

Fauci et al. (83) published a prospective study of 15 patients with systemic lymphomatoid granulomatosis. Of 13 patients treated with cyclophosphamide and prednisone, seven sustained complete remission (mean duration of remission, 5.2 ± 0.6 years). Two patients receiving only prednisone and six receiving cyclophosphamide and prednisone died. Six deaths were associated with biopsy-proven lymphoma; one was caused by a lymphomalike illness unproven by biopsy. The eighth death was caused by adenocarcinoma in a patient with lymphoma in remission. None of these patients received radiation therapy.

Chen et al. (43) reported their experience in 92 cases of LMG or centrofacial malignant lymphoma treated with radiation therapy. Twenty-five patients received combination chemotherapy, usually containing doxorubicin, cyclophosphamide, vincristine, and prednisone (CHOP) or other combinations, including CHOP or nitrogen mustard, vincristine, procarbazine, and prednisone (MOPP) in some patients. The nose was the most frequently involved site at initial presentation (85% of patients). Immunophenotyping in 36 patients showed T-cell lineage in 25 (69%) and B-cell in six (17%). The irradiation technique consisted of treating all involved and adjacent areas with doses of 30 to 75 Gy. Sixteen patients received neck irradiation (30 to 60 Gy). Daily fractions were 2 to 3 Gy in five weekly fractions. Actuarial survival rates were 59.5% at 5 years, 56.2% at 10 years, and 40.5% at 20 years. There was no significant difference in survival in patients receiving more or <50 Gy. A relapse in the midfacial region was noted in seven patients. Other relapse sites were lung and skin in three patients, para-aortic or inguinal lymph nodes in two patients, and brain in one. Survival of patients with recurrences was poor; 73% died within 8 months.

Hatta et al. (112) reviewed 18 patients (15 males and three females) with LMG (polymorphic reticulosis) (about 5.6% of patients with malignant head and neck tumors). Most of the 18 patients underwent both radiation therapy and chemotherapy (cyclophosphamide, vincristine, prednisone [COP], CHOP, methotrexate, leucovorin, doxorubicin, cyclophosphamide, vincristine, bleomycin, prednisone [MACOP-B]), but, since their disease had reached an advanced stage, three underwent radiation therapy only, three chemotherapy only, and one received no radical therapy. Of the 18 patients, 13 died of the disease; in six progress was confined to the local lesion. The 5-year cumulative survival rate was 15.7%. Fourteen autopsy studies revealed that tumor had invaded the liver (92.8%), lung (92.8%), and spleen (71.4%), and in all cases it was in leukemic patterns. Five cases were positive for ubiquitin carboxyl-terminal esterase L1 (ubiquitin thiolesterase) (UCHL-1) (CD45RO) and 10 cases were positive for lysozyme. All cases were positive for Ki-1 (CD30).

Sakata et al. (229) reported on 107 patients with stage I and II NHL of the head and neck treated with involved field radiation therapy for orbital, nasal, or paranasal lymphoma and extended field radiation for Waldeyer's ring or neck lymphoma (39 to 48 Gy). In the latter half of the study, adjuvant chemotherapy was administered. Of 107 patients, 95 achieved chemoradiation. Of the 12 patients who did not achieve chemoradiation, nine had nasal T-cell lymphoma (NTL) of the lethal midline granuloma (LMG-NTL) type. Only one patient who obtained chemoradiation relapsed in a previously irradiated area. LMG-NTL was the most significant prognostic factor on multivariate analysis (p <0.001). Older patients also experienced a higher relative risk than patients of 60 years of age or less (p = 0.0063). Dose of adriamycin reached borderline significance (p = 0.0600). Radiotherapy is excellent for obtaining local control of head and neck NHL and LMG-NTL.

Chloroma

Natural History
Chloroma (granulocytic sarcoma, myeloblastoma) is a solid extramedullary tumor composed of early myeloid precursors usually associated with acute myelocytic leukemia (41); the most common sites of presentation are in the orbit and other craniofacial bones. The name chloroma (from the Greek chloros, meaning green) derives from the green color of affected tissues resulting from the presence of myeloperoxidase. Because not all deposits exhibit the characteristic green tint, the term “granulocytic sarcoma” seems more appropriate.

Granulocytic sarcomas were identified in 3% of 478 patients with acute chronic granulocytic leukemia; they can be seen with other myeloproliferative disorders, including polycythemia vera, hypereosinophilia, and myeloid metaplasia (189,190). In the absence of acute leukemia, granulocytic sarcoma is usually an ominous sign, suggesting imminent conversion to acute myelocytic leukemia or blast crisis (190). As survival rates for myelogenous leukemias improve, the number of patients who relapse with chloromas is increasing (189).

Children are affected more often than adults. Of 33 patients with orbital chloromas reported by Zimmerman and Font (294), 75% were in their first decade of life. Chloromas are found more frequently in children with the M4 and M5 acute myeloid leukemia subtypes of the French-American-British Cooperative Group Classification and are also associated with the 8:21 translocation. Chloromas may appear during bone marrow remission before an increase in blasts is detected in the bone marrow, so they may herald relapse (189,190).

Clinical Presentation and Diagnostic Work-Up
Intraorbital (retrobulbar) chloroma causes progressive exophthalmos or temporal swelling. Central nervous system involvement causes both local pressure phenomena and generalized elevation of intracranial pressure with headaches, nausea, and vomiting (188).

Intracerebral chloromas may manifest as the rare central nervous system (parenchymal) involvement of acute nonlymphocytic leukemia. Woo et al. (291) believe that intracerebral chloromas represent reactivation of sanctuary deposits of leukemic cells in the central nervous system originated from an initial hematogenous spread.

All patients require complete hematologic and neurologic testing as is true for any patient with suspected leukemia. Open biopsy remains the best diagnostic tool. Plain radiographic findings consist mainly of localized bone destruction with predominantly lytic lesions and associated soft-tissue masses in orbital and periorbital chloromas. Intracranial chloromas may exhibit intermediate or high attenuation in unenhanced CT scans, with intense, uniform enhancement after intravenous administration of contrast material. Confusion with meningioma, hematoma, solitary metastasis, and lymphoma may occur on CT scans (215,244).

Gallium-67 scintigraphy was used to detect unsuspected leukemic infiltrates (165), but is not used any longer. Currently positron emission tomography (PET) scanning may be more useful for this purpose and to assess tumor response to therapy.

Radiation Therapy Techniques
Chloromas are extremely radiosensitive; however, the optimal dose of irradiation has not been established. Response rates of leukemic infiltrates have been reported with doses as low as 4 Gy, yet the need for higher doses up to 30 Gy in certain locations of extramedullary leukemic infiltrates is well recognized (189,190). Although the literature is limited regarding the maximum dose needed for treatment of chloromas, it appears that 30 Gy is the maximum required for local control. In our limited experience, there appears to be a relationship between the size of the chloroma and the total dose of irradiation required for control. The target volume is the tumor mass and an adequate margin (2 to 3 cm). Irradiation techniques depend on the location of the infiltrate. For superficial lesions, electron beam is recommended. Orbital chloroma may constitute a radiation therapy emergency because visual loss is possible if the patient is not treated promptly.

sabato 11 agosto 2012

52_01

Print Preview

Chapter 52

Breast: Stage Tis

David E. Wazer

Douglas W. Arthur

Noninvasive carcinoma of the breast (stage Tis) includes Paget's disease of the nipple and two histopathologic entities that are distinct in both their clinical presentation and biologic potential: Lobular carcinoma in situ (LCIS) and ductal carcinoma in situ (DCIS). As a result of the increase in the use of mammography, these three histopathologic entities comprise a larger percentage of all breast cancer cases seen today. There remains considerable controversy regarding the optimal treatment approach and, as a consequence, treatment recommendations range from observation to breast conservation therapy to mastectomy. It is, therefore, important to understand the distinguishing pathologic appearances, biologic characteristics and natural history of these three noninvasive breast disease entities in order to appropriately formulate coherent treatment recommendations.

Lobular Carcinoma in Situ

LCIS is characterized by multicentric breast involvement and consists of loose, discohesive epithelial cells that are large in size, variable in shape, and contain a normal cytoplasm to nucleus ratio (89). The extent of involvement of the lobular lumen ranges from simple filling to moderate-to-severe distention with extension into the adjacent extralobular ducts (124). As such, the lines of histologic delineation can become blurred between atypical ductal hyperplasia, LCIS, and, when ductal extension is seen, DCIS. This overlap of histologic morphology may complicate the interpretation of studies from different institutions (49,89,105,124).

LCIS has been reported to present with a multicentric distribution in up to 90% of mastectomy specimens, with bilateral involvement in 35% to 59% (89,105,124). LCIS cells are commonly estrogen-receptor positive, although overexpression of c-erbB-2 and p53 are uncommon (2,3,18,105). The loss of e-cadherin is often observed (2,64,129), and the absence of this adhesion molecule may explain the growth pattern seen with LCIS.

LCIS represents <15% of all noninvasive breast cancer (7,55,133). The majority of women are premenopausal at diagnosis, with an average age of 45 years (49,89,98). Risk factors for the development of LCIS correspond to those identified for invasive carcinoma (125). Because the male breast lacks lobular elements, this entity has not been described in men (49). As there are no clinical or mammographic indicators that are characteristic of LCIS, it is often detected as an incidental biopsy finding (89,105). In a minority of cases, LCIS can be detected with mammographic calcifications, but more commonly, calcifications are in adjacent tissue and are not histologically associated with LCIS (50,102,120). In excisional biopsy specimens, DCIS or invasive carcinoma are frequently identified even when LCIS is the sole histologic entity seen on core biopsy (24,47,74).

The presence of LCIS is considered a marker of increased risk for the subsequent development of invasive (usually ductal) carcinoma (7,49,55,98) that may be greatest for high-grade or more extensive lesions (86,89). This risk appears to be nearly equal for both breasts (22).

The question as to whether LCIS can serve as a direct precursor lesion to the subsequent development of invasive lobular carcinoma is unresolved. Some studies have suggested a clonal link of synchronously detected LCIS and invasive lobular carcinoma (60), whereas others have not (12). In an analysis of 182 patients with LCIS who were inadvertently enrolled on the National Surgical Adjuvant Breast and Project (NSABP) B-17 trial for DCIS and treated with lumpectomy only, there was a 14.4% in-breast tumor recurrence (IBTR) rate and a 7.8% contralateral breast tumor recurrence rate after a median follow-up of 12 years (44). Nine IBTR (5% of the total cohort) were invasive carcinoma and 17 (9% of the total cohort) were DCIS. Although the frequency of contralateral breast tumor recurrence rate was less than that of IBTR, the frequency of invasive contralateral breast tumor recurrence rate (5.6% of total cohort) was similar to invasive IBTR (5% of total cohort). Of note, all of the IBTR were documented to be at the site of the index lesion except for one, characterized as pure LCIS, that was found at a remote site.

Management for LCIS depends on whether it is associated with another malignancy (DCIS or invasive carcinoma) or if LCIS is the sole histologic diagnosis. Approximately 10% of early-stage breast cancers have an associated component of LCIS (1,81,103). The effect that the presence of LCIS has on the outcome of conservative management of early-stage breast cancer

P.1163


has only recently been evaluated. The most widely accepted treatment approach is to manage the breast according to the dominant malignant histology (DCIS or invasive carcinoma) and disregard the presence of LCIS. In such circumstances, it is not necessary to pursue additional surgery to obtain clear margins for LCIS (
1,12,81,103).

If LCIS is the sole histologic diagnosis, treatment recommendations range from conservative to radical. When first described as an entity, the significance of LCIS was unknown and mastectomy was often performed (46). The high frequency of contralateral breast involvement was subsequently used to justify contralateral biopsy and even bilateral mastectomy (46,98). Observational studies after wide local excision alone have led to a better understanding of the natural history of this condition, and a more conservative approach is now commonly practiced (7,49,55). In patients with LCIS as the sole histologic diagnosis, the most widely accepted clinical practice is close observation with regular physical examination and mammographic surveillance (7,44,49,55,133). There is no role for radiotherapy in the management of LCIS. The fact that LCIS commonly involves both breasts makes treatment with unilateral mastectomy both inadequate and illogical. Bilateral prophylactic mastectomy is likely excessive in all but those patients believed to be at highest risk: Young age, diffuse high-grade lesion, and significant family history. A less radical prophylactic approach in high-risk patients is to consider the use of tamoxifen. Tamoxifen has demonstrated efficacy in the prevention of invasive carcinoma and, in the context of LCIS, has been shown to reduce risk by 56% (40,128).

Paget's Disease

The clinical presentation of crusting and eczematous changes of the nipple–areola complex were first described in 1856. However, it was not until 1874 that the association with an underlying breast cancer was reported by Sir James Paget (90). Paget's disease of the nipple is characterized by the presence of Paget's cells that are located throughout the epidermis (76). Paget's cells are large and have hyperchromatic, round-to-oval nuclei with abundant amphophilic-to-clear cytoplasm. Mitoses are commonly seen, and the cells can be found in clusters or individually in the basal layers. The fact that Paget's disease is associated with an underlying malignancy in more than 95% of cases has generated discussion regarding the origin of these malignant cells. The epidermotropic theory appears to be the prevailing opinion with the belief that the disease originates from the underlying in situ or invasive disease. This is supported by histologic evidence of intraepithelial extension, immunohistochemical studies, and evidence suggesting that the epidermal keratinocytes release a motility factor, heregulin-α, that results in the chemotaxis of Paget's cells that migrate to the overlying nipple epidermis (30,34).

Paget's disease is a rare entity representing <5% of all breast cancer cases (65,100) and is typically diagnosed in the fifth or sixth decade. Synchronous bilateral and male Paget's disease have been reported (30,57,77).

Patients with Paget's disease describe itching and burning of the nipple and areola. There is a slow progression toward a crusting eczematoid appearance that can extend to the periareolar skin. If neglected, bleeding, pain, and ulceration can occur (100,130). Alternatively, Paget's disease can be asymptomatic and present as a pathologic finding after incidental surgical removal of the nipple–areolar complex (63). The differential diagnosis includes superficial spreading melanoma, pagetoid squamous cell carcinoma in situ, and clear cells of Toker (68,76). A palpable mass is detected in approximately 50% of patients at diagnosis; in more than 90% of cases this will be an invasive carcinoma. In contrast, if no palpable mass is detected, 66% to 86% will have an underlying DCIS. These associated malignancies are usually located centrally, although they can occur elsewhere in the breast (21,30,100). Mammographic findings are frequent in the presence of a palpable mass, but normal mammograms are reported in as many as 50% of cases (61,100).

At presentation, clinical evaluation includes bilateral breast examination, mammography, and biopsy to confirm the diagnosis of Paget's disease and to fully evaluate the extent of the associated malignancy. The prognosis does not dependent on the diagnosis of Paget's disease, but rather on the associated malignancy. Therefore, local treatment as well as systemic and regional nodal disease risk management should be based on the associated disease.

Management of Paget's disease continues to evolve. Mastectomy was employed in the past but this has been increasingly supplanted by breast-conserving treatment (8,91,136). The infrequent occurrence of this disease entity, the range of disease presentations (nipple involvement with/without an underlying mass and association with invasive vs. noninvasive disease), and the variable extent of surgical resection has made the evaluation of treatment options difficult. Small series have described results with various forms of breast-conserving treatment including wide local surgical resection alone, radiotherapy alone, and wide excision followed by whole-breast radiotherapy. Conservative surgery alone for Paget's disease appears to be inadequate, with reported local recurrence rates of 25% to 40% (32,39,48,70,94,126). The use of radiotherapy alone has been reported as achieving an 85% local control rate in a small series of patients with Paget's disease of the nipple who presented without an associated palpable mass (121). However, this approach has not been widely adopted because of the undefined histologic type and extent of the underlying disease leading to uncertainty in field design and total radiation dose.

The combination of limited surgical resection and postoperative radiotherapy appears to be the most practical breast-conserving approach. Two studies have evaluated the combined use of surgery and radiotherapy in Paget's disease of the nipple. The European Organization for Research and Treatment of Cancer (EORTC) Study 10873 was a multi-institutional registry trial that reported a 5-year local recurrence rate of 5.2% (16). In this study, a complete excision with tumor-free margins of the nipple–areolar complex and underlying breast tissue was followed by whole-breast radiotherapy. The median follow-up was 6.4 years, and the majority of these patients were found to have an underlying DCIS without a palpable mass. A separate study consisted of a seven-institution collaborative review of 36 patients with Paget's disease without a palpable mass or mammographic density (78,93). Patient follow-up was a median of 9.4 years. The extent of surgical resection varied as patients underwent complete (69%) or partial (25%) excision of the nipple–areolar complex and underlying breast tissue, with 6% reported as biopsy only. The final margin status was documented as negative in 56%, positive in 6%, and unknown in 39%. All received whole-breast irradiation and most received an additional boost dose to the tumor bed. The actuarial rate of local failure as the only site of first recurrence was 9% at 5 years and 13% at both 10 and 15 years. Two additional patients recurred in the treated breast simultaneously with regional and distant metastasis at 69 and 122 months. Despite the differences in clinical, pathologic, and treatment factors, statistical evaluation did not identify any factors that significantly predicted for risk of local recurrence.

Current data suggest that a combined-modality approach that conserves the breast is an appropriate alternative to mastectomy in properly selected patients with underlying noninvasive or invasive carcinoma of limited extent. As with any breast-conserving approach, patients with multicentric disease extension should be excluded. Surgical resection should include the nipple–areolar complex with microscopically clear margins

P.1164


surrounding both the Paget's disease and the associated malignancy. Whole-breast radiotherapy is delivered with standard techniques. Management of regional nodes and the risk of systemic disease is dictated by the associated malignancy.

Ductal Carcinoma in Situ

Clinical Presentation and Epidemiology

DCIS is a neoplastic process that is confined to the ductal system of the breast and lacks histologic evidence of invasion. These cells neither disrupt the basement membrane nor involve the surrounding breast stroma. This entity lacks the ability to metastasize and is confined to the breast (20,26,88,99). Axillary-node involvement is rare (0% to 5%) and most likely is associated with an undetected focus of invasive carcinoma (112). Risk factors for the development of DCIS are the same as those identified for invasive carcinoma (125) including family history, reproductive events such as delayed age of first live birth and nulliparity, history of benign breast biopsy, and dietary factors such as alcohol consumption. Before the use of screening mammography, DCIS typically presented as a palpable mass or nipple discharge. An invasive component commonly was found, and pure DCIS rarely was encountered. The widespread use of mammography now routinely detects DCIS <1 cm in diameter and results in breast cancer-free survival rates that approach 100% (112).

With the increased use of mammography and as pathologists began to recognize DCIS as a pathologic entity, the incidence of DCIS has markedly increased (73,113,114). The incidence of DCIS in the United States rose from 4,800 cases in 1983 to now more than 50,000 cases annually, representing a 10-fold increase in only 20 years (19). Of the 215,990 new breast cancers diagnosed in 2004, 59,390 were noninvasive, of which 85% were DCIS (66). Of these, 90% are nonpalpable (34). Studies have shown that the rate of screen-detected DCIS increases with age despite the fact that it accounts for a progressively smaller proportion of the total breast cancers detected (33). The rate of DCIS detection has been reported to increase from 0.56 per 1,000 mammograms among women aged 40 to 49 years to 1.07 per 1,000 mammograms among women aged 70 to 84 years (33).

Mammography

Ninety-five percent of new cases of DCIS present with mammographic abnormalities, of which microcalcifications are most typical (123). Noncalcified mammographic abnormalities make up the remaining findings, with asymmetric densities identified in 10%, dominant masses in 8%, and abnormal galactograms (performed for evaluation of nipple discharge) in 6%. Linear and branching calcifications frequently are associated with high-grade DCIS and necrosis, whereas fine and granular calcifications are associated more commonly with low-grade DCIS (Fig. 52.1, A and B) (31,58,96,135).

Initial evaluation should include magnification views that allow for complete characterization of mammographic findings and determination of the need for biopsy. The extent of the lesion as determined mammographically may be used as a guide for excision; however, the size typically is underestimated by 1 to 2 cm when compared with pathologic measurements (58,104). Ultrasonography, digital mammography, and magnetic resonance imaging all have the potential to be helpful in the management of DCIS but have yet to be proven as an acceptable substitute for mammography in screening (106). In cases that present with nipple discharge and a negative mammogram, galactography may be helpful in determining the likelihood of underlying DCIS versus papilloma (Fig. 52.1C) (92). Magnetic resonance imaging has the potential to refine clinical decision-making and surgical planning in select cases.

Pathology and Biology

The histologic diversity of DCIS can lead to difficulty in distinguishing it from other pathologic entities (88,99). The spectrum of DCIS extends from noncomedo, low-grade DCIS that can be similar in appearance to atypical ductal hyperplasia to comedo, high-grade DCIS. In addition, DCIS can extend into lobules, making it difficult to distinguish from LCIS (38). Traditionally, classification of DCIS has followed its architectural or morphologic appearance. The five subtypes of DCIS are comedo, solid, cribriform, micropapillary, and papillary (10,88,99) and it is common to encounter a mixture of subtypes within the same specimen (92). The characteristic features of each type are shown in Figure 52.2. Less common subtypes have been described and include apocrine, neuroendocrine, signet-cell cystic hypersecretory carcinoma, and clinging DCIS (71).

In 1997 a consensus conference committee was convened to reach an agreement on the pathologic classification of DCIS and the identification of specific features that may convey prognostic significance (26). Methods of processing and evaluating the pathologic specimen were also addressed. Rather than endorsing any specific classification system, the committee recommended and described features that should be documented for each case of DCIS, thus separating out important pathologic components and providing a comprehensive evaluation of the pathologic findings. These features include nuclear grade, presence of necrosis, polarization, and architectural pattern(s). The committee extended their recommendations to include margin status, lesion size, extent of microcalcifications, and correlation between specimen x-ray and mammographic findings. The DCIS Working Party of the EORTC arrived at similar conclusions and emphasized the importance of cytonuclear and architectural differentiation (97).

Three-dimensional examination and reconstruction techniques have resulted in a better understanding of the enormously complex structure of the mammary duct–lobular system and the patterns by which DCIS can spread within the breast (51,84,85) (Fig. 52.3). Knowledge of the anatomy and distribution of DCIS within the mammary ductal tree can be useful in selecting patients for breast conservation and assuring maximal surgical clearance of the lesion while preserving an acceptable cosmetic result. For example, Ohtake et al. (84,85) studied the duct–lobular system with computer graphic reconstruction and found that the breast consists of 16 to 24 duct–lobular systems, each culminating in a corresponding collecting duct at the nipple. They also identified ductal anastomoses that established a connection between the various ductal–lobular units and provided a potential pathway for tumor extension and subsequent diffuse involvement (84,85). Their proposed model for the development of widespread intraductal tumor extension within the breast is seen in Figure 52.4.

Faverly et al. (37) have described DCIS growth pattern within the ductal tree and the implications for surgical excision. The growth patterns documented include unicentric (one area only), multicentric (two distinct areas separated by more than 4 cm), continuous (extension along ductal system without gaps), and discontinuous or multifocal (two or more areas separated by <4 cm). They found that in mammographically detected DCIS, a multicentric growth pattern was rare (less than 2%), with most cases showing an even distribution between discontinuous and continuous growth patterns. Of cases with a discontinuous growth pattern, 63% had foci separated by gaps that measured <5 mm, 83% had foci separated by <10 mm, and only 8% had foci separated by >10 mm. There was a correlation between differentiation and growth pattern such that 90% of poorly

P.1165


differentiated DCIS showed a continuous growth pattern, whereas 70% of well-differentiated DCIS had a discontinuous growth pattern. Based on these findings, the authors concluded that a 1-cm margin of normal tissue around the lesion would lead to complete surgical clearance of histologically evident DCIS in 90% of cases.

DCIS is a precursor lesion to invasive ductal carcinoma and exists along an evolutionary continuum that starts with benign breast tissue and ends with an invasive breast carcinoma (4). This concept has been validated in several ways. For years, pathologists have recognized and documented confirmation of a histologic progression from benign breast cells to invasive breast cancer. The evolutionary concept is further supported by the recognized association between the presence of DCIS and the subsequent increased risk of developing an invasive breast cancer (19,95,122). In some series, a 10-fold risk of developing an invasive lesion has been reported. Most importantly, the presence of shared identical genetic abnormalities between DCIS and synchronous invasive breast cancer demonstrates a clonal relationship of biologic progression (19,83,95,122). The biologic evolution from benign breast cells to invasive breast cancer occurs through highly diverse genetic mechanisms.

Genetic and molecular differences have been documented that differentiate DCIS from normal breast tissue. Genetic alterations have been evaluated with an analysis of loss of heterozygosity that has demonstrated gain or loss of multiple loci (9,36,83,95,122).

P.1166


Loss of heterozygosity is not seen in normal breast tissue. The frequency of loss of heterozygosity correlates with histologic progression of breast tissue from benign to malignant. Loss of heterozygosity is seen in approximately 50% of atypical ductal hyperplasia. Among specimens harvested from cancerous breasts, 77% of noncomedo and 80% of comedo DCIS lesions share loss of heterozygosity with the synchronous invasive lesion in at least one locus (
83).

Molecular markers have been studied in DCIS and are found to have a heterogeneous distribution of expression (19). The estrogen receptor is present in 70% of DCIS but the rate of expression is higher in low-grade lesions (90%) than in high-grade lesions (25%). This association with histologic grade is reversed for the rate of overexpression of HER2/neu proto-oncogene and the p53 tumor suppression gene. Approximately 50% of all DCIS lesions have overexpression of HER2/neu, and in 25% the p53 tumor suppressor gene is also detected. Both of these molecular markers are noted in <20% of low-grade lesions but are present in approximately two thirds of high-grade lesions.

Alterations in the surrounding breast parenchyma may also be seen with DCIS. High-grade DCIS, in particular, has been associated with the breakdown of the myoepithelial cell layer and basement membrane surrounding the ductal lumen (28), proliferation of fibroblasts, lymphocyte infiltration, and angiogenesis in the surrounding stromal tissues (52,53). Whether these stromal changes reflect important steps that facilitate primary tumor transformation or secondary alterations in response to ductal epithelium that is being transformed is unknown. Quantitative changes in the expression of genes related to cell motility, adhesion, and extracellular-matrix composition, all of which may be related to the acquisition of invasiveness, occur as DCIS evolves into invasive carcinoma (5).

P.1167


Data suggest that DCIS represents a stage in the development of breast cancer in which most of the molecular changes that characterize invasive breast cancer are already present, although the lesion has not yet assumed a fully malignant phenotype. A final set of events, which probably includes gain of function by malignant cells and loss of function and integrity by surrounding normal tissues, is associated with the transition from a preinvasive DCIS lesion to invasive cancer. Most, if not all, clinically relevant features of breast cancer, such as hormone-receptor status, the level of oncogene expression, and histologic grade, are probably determined by the time DCIS has evolved (17,54,72,131).

An occult microinvasive tumor (one that does not exceed 0.1 cm in diameter) may be seen with some cases of DCIS. Such cases are classified as microinvasive breast cancer (115) and are generally treated according to the guidelines for invasive disease. Occult microinvasive tumors are most common in patients with DCIS lesions that are >2.5 cm in diameter (69), those presenting with palpable masses or nipple discharge, and those with high-grade DCIS or comedonecrosis (92,107).

Natural History of DCIS

The overall incidence of DCIS in the general population is unclear. In an attempt to address this question, a small number of autopsy studies have been reported. One series examined 185 randomly selected breasts from 101 women in which a subgross sampling technique was used (6) and one or more foci of DCIS were found in 6% of cases. A review of seven autopsy series of women not known to have breast cancer during life showed a median prevalence of DCIS of 8.9% (range, 0% to 14.7%) (132). The fact that some autopsy series document a greater incidence of DCIS in asymptomatic women than most clinical series suggests either the possibility that DCIS is either underdiagnosed or that many cases are not clinically significant.

A primary consideration in the natural history of DCIS is the risk of progression to invasive carcinoma. The published evidence on the clinical course of untreated DCIS is sparse because it has been recognized as a distinct entity for only a relatively brief period, having been considered rare before the widespread use of mammography and having been treated most frequently by mastectomy. Those cases for which long-term follow-up data are available were grossly palpable DCIS, a form that may not be equivalent to the mammographic DCIS that is seen more commonly today. The few published long-term follow-up studies of DCIS after only biopsy document an overall incidence of subsequent invasive carcinoma of more than 36% (13,25,80,101). Most of these subsequent malignancies occur within 10 years, although as many as one-third may develop after 15 years (13,101).

Women with DCIS in one breast are at risk for a second tumor (either invasive or in situ) in the contralateral breast (56); the rate at which such tumors develop is similar to that among women with primary invasive breast cancer, approximately 0.5% to 1% per year.

DCIS is a part of the breast/ovarian cancer syndromes defined by BRCA1 and BRCA2, with mutation rates similar to those found for invasive breast cancer (23). These findings suggest that patients with DCIS with an appropriate personal or family history of breast and/or ovarian cancer should be screened and followed according to the same high-risk protocols as developed for invasive breast cancer.

Treatment Options for DCIS

Prognostic Factors and Their Interpretation

The goal of treatment with DCIS is prevention of local recurrence, with particular emphasis on the prevention of invasive breast cancer. Treatment decisions are largely based on information provided by mammography and, most especially, pathologic evaluation of the biopsy specimen. As such, in the consideration of treatment options it is important for the clinician to

P.1168


be aware of some of the technical limitations associated with the clinical and histopathologic assessment of DCIS.

Studies performed during the past two decades clearly have suggested that DCIS is not a single disease. Rather, this term encompasses a diverse group of lesions that differ with regard to their clinical presentation, mammographic features, extent and distribution within the breast, histologic characteristics, and biologic markers. Moreover, clinical follow-up studies have indicated that these lesions vary in their propensity to recur or progress to invasive breast cancer. As a consequence, a significant proportion of patients diagnosed with DCIS can be treated adequately with breast-conserving therapy (i.e., excision with or without radiation therapy). Which patients with DCIS can be treated safely with excision alone and which patients require radiation therapy after excision is a pressing clinical question. Attempts to resolve this issue have focused on the identification of risk factors for local recurrence after breast-conservation therapy for DCIS. With three exceptions (the prospectively randomized NSABP B-17, the EORTC 10853, and the United Kingdom Coordinating Committee on Cancer Research [UKCCCR] trials), all such studies have been retrospective in design. Nonetheless, a number of factors have been identified that may be important in defining local failure risk. These include symptomatic presentation (15,87,116), lesion size (108,116), histopathologic subtype (15), nuclear/cytologic grade (87,116,119), central necrosis (87,116,119), margin status (109,116,117), and patient age (15,42,117,127).

The relative importance of any histopathologic factor in predicting the probability of local recurrence and, in turn, selecting the appropriate therapeutic option for a given patient is unclear. This is partly the result of the inherent difficulty associated with the establishment of standardized and reproducible systems of pathologic classification, including such apparently straightforward assessments of grade, margin width, and lesion size.

Recent efforts to classify DCIS have been based primarily on the nuclear grade of the lesion and/or the presence or absence of necrosis. A number of studies have shown that there is an association between high nuclear grade and/or necrosis and the risk of local recurrence and progression to invasion (87,116,119). Although the criteria for histologic grading systems have been published, there are limited data regarding the ability of pathologists to apply them in a reproducible manner.

Several studies have shown that the status of the microscopic margins appears to be important in predicting the likelihood of recurrence in the breast for patients with both invasive breast cancer and DCIS treated with breast-conserving therapy (109,116,117). However, there are numerous technical problems in the evaluation of margins of breast-excision specimens. First, if a specimen is removed in more than one fragment, the margins cannot be evaluated. Second, there is no standardized method for sampling or reporting margins, and this process is subject to sampling error. Finally, it is often difficult to provide an accurate assessment of the margin width for patients who undergo a re-excision as the initial biopsy site can be eccentrically located in the surgical specimen.

Most DCIS lesions present as a nonpalpable, grossly inapparent mammographic abnormality, which can make accurate determination of the size or extent of the lesion difficult (Fig. 52.5). The two modalities available to assess the size of the lesion are mammography and pathologic examination. Mammography frequently will underestimate the pathologic extent of DCIS, particularly for well-differentiated lesions in which substantial areas of the tumor may not contain microcalcifications. Pathologic assessment of lesion size also can be difficult. Macroscopic examination of a specimen containing DCIS rarely reveals a grossly evident tumor that can be measured. Therefore, the assessment of the size of the lesion must be estimated from histologic sections.

Mastectomy for DCIS

Mastectomy was the standard treatment of DCIS through the first four decades of its recognition as a distinct histopathologic entity. Mastectomy is a highly effective treatment for DCIS, with a locoregional control rate of 96% to 100% and cancer-specific mortality rates of 4% or less (111). No randomized study has compared mastectomy with breast-conservation treatment for DCIS. Therefore, the relative outcomes for mastectomy and breast-conservation treatment can be estimated only by reviewing nonrandomized, retrospective studies. Local treatment failure after mastectomy (111) may occur because of unrecognized invasive carcinoma that results in local recurrence or distant metastasis, or it may be the result of incomplete removal of breast tissue with the subsequent formation of a new primary tumor.

Data from some surgical trials (45) and large treatment registries (35) suggest that the rates of local or regional recurrence are significantly lower after mastectomy than after breast-conserving surgery, but there have been no significant differences in overall survival. Metastatic breast cancer can follow the recurrence of an invasive tumor or the development of cancer in the contralateral breast. However, death related to breast cancer within 10 years after the diagnosis of DCIS occurs in only 1% to 2% of all patients, irrespective of whether mastectomy or breast-conserving surgery was performed (35).

Breast Conservation for DCIS

Three prospective randomized studies of excision only versus excision plus breast irradiation for DCIS have been performed, and all have shown that the rate of local recurrence was reduced with the addition of radiation (Table 52.1). The NSABP B-17 trial (41,43) consisted of 818 patients who were stratified by age (49 years of age or younger vs. older than 49 years), DCIS versus DCIS plus LCIS, method of detection, and whether an axillary dissection was performed. Tumor size was determined by mammogram, gross pathologic measurement, or clinical examination. Of the patients enrolled, 83% had nonpalpable tumors. The 12-year rate of local recurrence was 15.7% with radiation and 31.7% without radiation (p <.000005) (Fig. 52.6). The average annual incidence rates of all ipsilateral breast tumor recurrences, ipsilateral noninvasive recurrences, and ipsilateral invasive recurrences were reduced with breast irradiation by 59%, 47%, and 71%, respectively. An analysis of clinical

P.1169


variables showed that microcalcifications extending beyond a maximum dimension of >1 cm were associated with an elevated risk of breast recurrence. A central pathology review was performed, including a multivariate analysis of histopathologic variables (
Table 52.2), that revealed only moderate/marked comedonecrosis as being significantly associated with local failure risk. Margin status (free vs. unknown/involved) was of borderline significance.

The EORTC 10853 trial (14,67) randomly allocated 1,010 patients with 5 cm or smaller DCIS and negative margins to excision versus excision plus breast irradiation. Lesions were nonpalpable in 79% of patients, and the mean maximal tumor diameter was approximately 2 cm. The 10-year rate of local recurrence was 15% for patients treated with radiation, as compared with 25% for patients treated without radiation (p <.0001). At a median follow-up of 10.2 years, radiation therapy resulted in risk reduction for both invasive and noninvasive breast relapse of 42%. As with the NSABP B-17 study, a central pathology review was performed (14,15). In a multivariate analysis (Table 52.3), factors associated with an increased risk of local recurrence were age 40 years or younger, clinically symptomatic presentation (nipple discharge or palpable mass), intermediate or poorly differentiated DCIS, solid and cribriform histologic growth pattern, involved or uncertain margins, and treatment by local excision alone. The risk of invasive recurrence was not related to histologic type of DCIS, but the risk of distant metastasis was significantly higher in poorly differentiated DCIS compared with well-differentiated DCIS.

The EORTC 10853 trial did not allow the identification of an appropriate margin width for treatment with or without radiotherapy because the eligibility criteria did not require reporting of the margin status. Nonetheless, the central review of cases did provide some information regarding the relative importance of surgical margin as related to local failure risk. A recurrence rate of 24% at 4 years was observed in cases with close/involved margins after excision alone. Radiotherapy was not adequate to compensate for involved margins because even with the application of irradiation the recurrence rate was 20% in this group. These data and others (108,109,116,117) are strongly suggestive that obtaining a microscopic complete excision is essential for optimal local control in breast-conserving therapy for DCIS. Of further note, even in the group of DCIS cases for which margins could be considered optimal (i.e., those patients who underwent a surgical re-excision in which no residual DCIS was found), a 4-year local recurrence rate of 18% was observed when these patients were treated with surgery alone (15).

The UKCCCR DCIS Working group has also conducted a randomized trial investigating the role of adjuvant radiotherapy (59). With a 2 × 2 factorial protocol design, the aim of this study was to compare excision alone versus excision plus tamoxifen versus excision plus radiotherapy versus excision plus radiotherapy and tamoxifen. Tamoxifen was prescribed as 20 mg per day and radiotherapy was delivered through whole-breast tangential fields to a total dose of 50 Gy. Boost was not recommended. A total of 1,030 patients were enrolled. When reported with 4.38 year follow-up, the crude incidence of local recurrence was 14% of the patients who were treated with excision only and 6% when the excision was followed by radiotherapy. The addition of tamoxifen offered no benefit toward overall ipsilateral local control when administered in addition to radiotherapy; however, tamoxifen did appear to reduce the ipsilateral recurrence rate of DCIS (but not invasive carcinoma) in the absence of radiotherapy (59).

Subgroup analyses from randomized trials have demonstrated that the absolute benefits of radiotherapy are greater in women at increased risk for tumor recurrence, such as women with involved surgical margins (identified on retrospective pathologic review), younger women, and those with tumors that have high-grade or comedonecrotic features (14,15,41,43). However, radiotherapy lowers the incidence of recurrence among all subgroups, regardless of the baseline risk.

Patient age is an important prognostic variable for local recurrence after breast conservation for DCIS (15,42,117,127). In younger patients, DCIS more frequently contains adverse prognostic pathologic features and extends over a greater distance in the breast than in older patients (127). In series with

P.1170


adequate follow-up, younger patients treated with lumpectomy and radiation therapy had a significantly higher rate of local recurrence than older patients, especially for invasive local recurrences (
127). Some studies have suggested that careful attention to margin status and excising larger volumes of tissue can reduce this difference substantially (117,127). No available data show that younger patients have better long-term cancer-free survival rates if treated by mastectomy rather than lumpectomy and radiation therapy. Successful treatment of younger patients with DCIS with lumpectomy and radiation therapy requires careful attention to patient evaluation, selection, and surgical technique. When this is done, age at diagnosis should not be a contraindication to breast-conserving therapy.

A number of recent studies have attempted to identify and treat patients with highly selected favorable tumor characteristics with excision alone (i.e., without whole-breast irradiation) and report 10-year local failure rates of 3% to 25% (109,111). A scoring system has been proposed (108) using histopathologic features including tumor size, grade, and margin width in an attempt to stratify patients according to local failure risk after excision plus or minus whole-breast irradiation. Each variable was assigned a score of 1 to 3, and the sum total defined the Van Nuys Prognostic Index. Although appealingly simple, this scheme (108) is drawn from the retrospective analysis of a patient cohort in which there exist a number of methodologic shortcomings and it has not been independently validated (29).

Wong et al. (134) performed a prospective study that attempted to identify patients with “low-risk” DCIS who can be spared whole-breast radiation therapy. This trial enrolled 158 patients with lesions that were mostly grade 1 or 2 and with a mammographic extent of ≤2.5 cm who were treated with wide excision, with final margins of ≥1 cm or a re-excision without

P.1171


residual DCIS. Tamoxifen was not permitted. The median age was 51 years and the median follow-up was 40 months. The rate of ipsilateral local recurrence was 2.4% per patient-year, corresponding to a 5-year rate of 12%. Nine patients (69%) experienced recurrence of DCIS and four (31%) experienced recurrence with invasive carcinoma. These data provide prospective evidence that, despite margins of >1 cm, the local recurrence rate is substantial even in patients with small, grade 1 or 2 DCIS following treatment with wide excision alone.

Presently, the Radiation Therapy Oncology Group is conducting a prospective randomized trial to further assess the need for radiotherapy in low-risk DCIS. Following lumpectomy with ≥3 mm clear margins of resection, patients are stratified according to age (<50 vs. ≥50 years), tumor size (≤1 vs. >1 to 2.5 cm), margin status (negative re-excision vs. 3 to 9 vs. ≥10 mm), grade, and the use of tamoxifen (at the discretion of the managing physician). Following stratification, patients are randomized to whole-breast irradiation versus observation. The NSABP and Radiation Therapy Oncology Group have jointly launched a phase III accelerated partial-breast irradiation trial that randomly allocates patients between standard whole-breast irradiation following lumpectomy versus accelerated partial-breast irradiation to determine if in-breast control rates are comparable. As the in-breast failure patterns for DCIS suggest that treatment directed to the primary lesion plus a 2-cm margin should achieve local control rates that equate to whole-breast treatment approaches, patients with pure DCIS or DCIS and LCIS will be eligible for stratified randomization.

Follow-Up and Management of Recurrence

Ipsilateral tumor recurrences in patients with DCIS are usually detected on surveillance mammography, although one-quarter may be detected on the basis of changes on physical examination of the breast or chest wall (75,118). For this reason, patients should be scheduled for a baseline mammogram 6 to 12 months after initial therapy and at least annually thereafter. Distant breast cancer metastases in the absence of regional recurrence are unusual. Local recurrences after breast-conserving surgery and radiotherapy are generally treated with mastectomy. Selected patients with local recurrences who have not previously received radiotherapy may be candidates for local excision and radiotherapy. The clinical outcome of ipsilateral tumor recurrence is governed by the nature of the recurrence. Patients with recurrent DCIS have an excellent prognosis, with less than a 1% risk of further recurrence after salvage mastectomy. Patients with invasive recurrence after breast-conserving surgery for DCIS have a prognosis similar to those with early-stage breast cancer, with a 15% to 20% risk of metastatic recurrence at 8 years (118).

The Role of Tamoxifen for DCIS

The NSABP B-24 trial (42) compared excision plus radiotherapy to excision, radiotherapy, and tamoxifen. Patients who received tamoxifen had a decreased incidence of breast cancer events (invasive or noninvasive ipsilateral or contralateral breast cancer) compared with patients who did not receive tamoxifen (8.2% vs. 13.4% at 5 years, respectively; p = .0009), but no survival benefit was found. Tamoxifen therapy resulted in a 44% reduction in the risk of subsequent invasive tumor recurrence but had no significant effect on ipsilateral noninvasive breast recurrence (Table 52.4). Positive tumor margins were significantly associated with breast recurrence, and tamoxifen reduced ipsilateral breast failure by 22% with negative margins and 44% in cases with positive or unknown margins.

In contrast to the findings of the NSABP B-24 trial, the UKCCCR trial found that tamoxifen had no effect in reducing local recurrence rate when combined with whole-breast radiation therapy (Table 52.4). When used as single agent without radiation therapy after lumpectomy, tamoxifen had no effect on the incidence of invasive recurrence but did show a statistically significant reduction in the risk of DCIS recurrence (10% vs. 6%; p = .03) (59). As such, the role of tamoxifen for DCIS in the absence of whole-breast radiotherapy remains to be defined.

Because DCIS is a precursor to invasive breast cancer and shares many biologic features of invasive carcinoma, it is increasingly recognized as a target for preventive measures. In the largest trials of the prevention of primary breast cancer among women at high risk for breast cancer by virtue of age, family history, or prior benign breast disease, tamoxifen reduced the risk of DCIS by 50% to 70% (27,40).

A Decision Tree for DCIS

The management of DCIS requires the coordinated, multidisciplinary interaction of radiologists, surgeons, pathologists, and oncologists. Patients are first assessed to determine if they are candidates for breast-conserving surgery. Women with multicentric DCIS, as defined by the presence of two or more tumors in separate quadrants of the breast, and those with extensive or diffuse DCIS or suspicious-appearing microcalcifications throughout the breast are candidates for mastectomy, as are women in whom negative margins or acceptable cosmesis cannot be achieved with the use of breast-conserving surgery. Some women may prefer mastectomy to breast conservation in order to minimize the chance of ipsilateral recurrence or for other reasons. At present, there is no established role for the use of magnetic resonance imaging in screening patients for DCIS in determining whether breast-conserving surgery is an option.

Patients deemed to be appropriate candidates for breast conservation require complete surgical excision of the affected area. The extent of DCIS in the breast and the existing margin determine the likelihood of identifying residual disease on reexcision. Nearly half of patients with margins that are <1 mm have residual DCIS on re-excision (82). However, the optimal margin width for the management of DCIS is not known. At a minimum, there should be no tumor at the margin.

P.1172


Neither dissection of axillary lymph nodes nor mapping of sentinel lymph nodes is routinely warranted in patients with DCIS because of the very low incidence of axillary metastases (110). Three to 13% of patients with DCIS, and a slightly greater percentage with DCIS and microinvasion, have isolated tumor cells in sentinel axillary lymph nodes (62). The prognostic significance of these cells is not clear. Clinical experience suggests that patients have a much better outcome than would be predicted by such rates of nodal metastases, and most instances represent micrometastases of unclear metastatic potential. However, sentinel lymph node mapping may be used in selected patients with a higher likelihood of occult invasive cancer—those with extensive, high-grade DCIS or palpable masses—and those undergoing mastectomy as sentinel node mapping cannot be performed afterward if invasive tumor is identified (79).

After breast-conserving surgery, radiotherapy is administered using tangential fields to the whole breast with a standard dose of 45 to 50 Gy delivered in daily fractions of 180 to 200 cGy. On the basis of extrapolation from data on the treatment of invasive breast cancer (11), a radiation boost to the tumor bed may be added to whole-breast treatment, particularly for women with close surgical margins, although the benefit of a boost in the management of DCIS is not established. There is no role for postmastectomy or nodal irradiation in the treatment of DCIS.

It is not yet possible to prospectively identify women who are at sufficiently low risk that radiotherapy may not be of some clinical advantage in preventing recurrences. After discussing the various options, patients may elect not to receive radiation treatment, but they must understand and accept the increased risk of recurrence that this choice probably entails.

In summary, despite considerable advances in our clinical knowledge base, the answer to the question “when should radiotherapy be used for DCIS?” remains complex and surrounded by considerable controversy. Two fundamental considerations must be emphasized:

  • A primary goal of breast-conserving therapy for DCIS is to achieve the best possible cosmetic outcome. Attempts to obtain wide surgical margins through deforming, large-volume breast excisions represent cosmetic failures and defeat the purpose of breast conservation.
  • Breast irradiation reduces the risk of subsequent invasive or noninvasive carcinoma in the treated breast and thus reduces the risk of the ultimate cosmetic failure— mastectomy.

According to prospectively randomized trials of breast-conserving therapy for DCIS, radiotherapy reduces subsequent breast recurrence in all patient groups irrespective of prognostic risk factors. That is not to say, however, that radiotherapy must be used for all patients with DCIS. In all cases, a realistic and balanced discussion of the relative risks and benefits of treatment options should be presented to the patient. Reasonable estimates of breast recurrence during the ensuing decade with or without radiotherapy are available based on level I evidence from prospective clinical trials. A decision tree to assist in the selection of treatment options is presented in Table 52.5.