Fax: (011) 353-1-4974886
Should extrapulmonary small cell cancer be managed like small cell lung cancer?
Article first published online: 5 JAN 2010
Copyright © 2010 American Cancer Society
Volume 116, Issue 4, pages 888–895, 15 February 2010
How to Cite
Brennan, S. M., Gregory, D. L., Stillie, A., Herschtal, A., Mac Manus, M. and Ball, D. L. (2010), Should extrapulmonary small cell cancer be managed like small cell lung cancer?. Cancer, 116: 888–895. doi: 10.1002/cncr.24858
- Issue published online: 2 FEB 2010
- Article first published online: 5 JAN 2010
- Manuscript Revised: 30 JUN 2009
- Manuscript Accepted: 30 JUN 2009
- Manuscript Received: 30 APR 2009
- extrapulmonary small cell carcinoma;
- prophylactic cranial irradiation;
- small cell lung cancer
The aim of this study was to determine if extrapulmonary small cell carcinomas (EPSCC) should be managed using protocols similar to those for small cell lung cancer (SCLC).
Treatment strategies, survival, patterns of failure, and prognostic factors for patients with EPSCC were analyzed retrospectively at a large cancer center. SCLC was excluded by thoracic computed tomography (75%) or chest radiography (25%).
Of 120 eligible patients, 70% had limited disease (LD). Treatment modalities included chemotherapy (n = 82; 68%), radiotherapy (RT) (n = 80; 67%), and surgery (n = 41, 34%). The median survival for patients with LD and extensive disease was 1.4 years and 0.7 years, respectively. Gynecologic (n = 31) and gastrointestinal (n = 28) were the most common primary tumor sites. Gynecologic and head and neck primary tumor sites had better 1-year survival than other sites (P = .019 and 0.005, respectively). Brain metastasis was the site of first distant failure in 4.1% of patients versus 35% for soft tissue metastases. The lifetime risk of brain metastasis was 13%. Definitive RT (P = .004), LD (P = .028), and prophylactic cranial irradiation (PCI) (P = .022) were found to be positive prognostic factors and weight loss (P < .001) was a negative prognostic factor on multivariate analysis.
Patients with EPSCC usually experienced short survival, often with early distant metastasis. Although PCI was associated with improved overall survival, brain metastasis was less frequent than in patients with SCLC, and therefore the potential benefit of PCI was less than in patients with SCLC. Definitive chemoradiotherapy was associated with better outcomes and should be delivered whenever feasible. Cancer 2010. © 2010 American Cancer Society.
Extrapulmonary small cell carcinoma (EPSCC) was initially described as a distinct clinicopathologic disease entity in 1930 by Duguid and Kennedy.1 It constitutes a rare, aggressive disease with a reported incidence of 0.1% to 0.4% in North America2 and represents up to 5% of all cases of small cell carcinoma.3 The most commonly reported primary sites of disease are the gastrointestinal tract, genitourinary system, and EPSCC of unknown primary tumor.4 This rare, heterogeneous carcinoma poses a significant diagnostic and therapeutic challenge and is characterized by early dissemination and poor prognosis.
To the best of our knowledge, no prospective trials exist for this disease, and treatment recommendations are derived from small single-institution experiences4-8 or extrapolated from studies of small cell lung cancer (SCLC). Although SCLC is often considered to be analogous to EPSCC, the existence of significant biological differences between these diseases could lead to changes in emphasis when treatment strategies are considered. Unfortunately, reports in the literature are comprised of either small series of well-characterized patients with single primary subsites9-11 or, alternatively, are comprised of larger series of less well-characterized patients with multiple primary subsites, making it difficult to draw useful conclusions regarding the natural history of EPSCC compared with SCLC. Most of these series indicate wide variation in median survival between different primary sites,4, 7, 12, 13 suggesting that extrapolation from a single common site such as lung may not accurately predict the behavior of small cell carcinoma at other sites.
The use of multimodality therapy has been shown to impact upon survival in SCLC, with the addition of thoracic radiotherapy (RT) to chemotherapy conferring a 5% survival advantage and yielding complete response rates of 50% to 85%.14 To the best of our knowledge, the optimal management in EPSCC has yet to be elucidated, and prognostic risk factors have not been clearly established for this disease.
The incidence of cerebral metastases in SCLC ranges from 20% to 30% in limited disease15 (LD) and up to 40% in extensive disease (ED).16 Prophylactic cranial irradiation (PCI) is routinely offered because it confers a 5.4% survival benefit at 3 years.15 The role of PCI in EPSCC is controversial. In some smaller series, the incidence of brain metastases appears to be lower than in pulmonary SCLC and, therefore, PCI is not uniformly recommended.5, 6, 10 Other authors11, 17 have highlighted the need for larger series to further investigate the incidence of brain metastases in this population and the potential role for PCI.
The purpose of this study was to examine this uncommon cancer in the setting of a tertiary referral center in an attempt to accurately reflect modern histologic and radiologic diagnostic patterns and multimodality treatment approaches. The prognostic impact of various patient and treatment factors on survival and recurrence patterns was examined. We hoped to determine whether it is reasonable to apply management strategies designed for SCLC to patients with EPSCC.
MATERIALS AND METHODS
This study was approved by the institutional ethics committee. The study cohort was identified by searching the institutional database for patients diagnosed with EPSCC between 1985 and 2007 using ICD 10 (International Classification of Diseases) morphology code M80413. To be eligible for study inclusion, all patients were required to be at least 18 years of age, without evidence of a primary pulmonary small cell carcinoma on chest radiograph or computed tomography (CT) of the thorax. Patients with well differentiated neurendocrine carcinomas or Merkel cell carcinomas were excluded. Patient characteristics, staging investigations, treatment, response and outcome data were entered into a study specific database. Patients were allocated to LD or ED by analogy with the 2 stage Veterans Affairs Lung Study Group staging system.18 The LD category included patients with local or locoregional disease encompassable within a tolerable RT portal. For patients with unknown primaries, limited stage included patients with disease encompassable within a tolerable RT portal. The categorization of patients as limited or extensive stage was based on the review of the clinical notes and radiology reports. Soft tissue metastases included all sites of distant metastases other than brain or bone metastases.
The Kaplan-Meier product limit method was used to calculate the rates of disease recurrence and survival. All recurrence-free durations were measured from the date of diagnosis until documentation of recurrence. Locoregional recurrence was censored by distant recurrence, as well as by death from any cause. Distant recurrence was censored by death from any cause. Univariate and multivariate analyses were conducted to identify prognostic risk factors for survival. All patients were included in overall survival analyses, but only patients with LD treated with potentially curative (definitive) treatment with surgery, RT, or chemotherapy, were included in the analyses of recurrence-free survival.
Patient Characteristics and Primary Disease Sites
There were 120 patients with a diagnosis of EPSCC (Table 1). Eighty-four patients (70%) had LD. The most frequent disease region was the female genital tract (gynecologic primary sites) (n = 31; 26%), followed by gastrointestinal tract (n = 28; 23%), genitourinary tract (n = 23; 19%), head and neck (n = 19; 16%), unknown primary site (n = 15; 13%), and other sites (n = 4; 3%). Genitourinary sites included bladder and prostate and excluded female genital tract tumors, which were considered separately due to the large number of these tumors. An independent pathologic review was performed at the time of diagnosis in 66 patients (55%) to confirm the diagnosis of EPSCC. All patients had a chest radiograph, and 90 (75%) had a CT scan of the thorax (Table 2).
|No., n=120||Percentage (100%)|
|ECOG performance status||0||36||30|
|Staging Investigation||No., n=120||Percentage (100%)|
|CT scan of thorax||90||75|
|CT scan of primary tumor site||105||88|
|MRI of primary tumor site||18||15|
|CT scan of brain||22||18|
|MRI of brain||7||6|
All 15 patients with unknown primary tumors (100%) underwent a CT scan that included the thorax. In addition, 3 of the 15 patients (20%) underwent positron emission tomography (PET)/CT. For all patients, the median follow-up was 1 year (range, 14 days–20 years). One patient died 14 days after diagnosis before receiving treatment. For the 20 surviving patients, the median follow-up was 5 years and 1 month (range, 12 months–20 years). Of the total group, 19% were current smokers, 32% were former smokers, and 29% were nonsmokers. Smoking status was unknown in 20%. The most common primary tumor subsites were cervix (n = 24; 77% of gynecologic tumors), bladder (n = 15; 65% of genitourinary tumors), esophagus (n = 20; 71% of gastrointestinal tumors), and parotid (n = 5; 26% of head and neck tumors). Small cell carcinoma of the cervix was the most common site overall (n = 24; 20% of the entire group of 120 patients). The proportion of patients with LD varied and was much higher in gynecologic (84%) and head and neck (84%) tumor sites, compared with the gastrointestinal tract (57%), genitourinary tract (65%), and unknown primary (47%) tumor sites. The proportion of patients with >10% weight loss was much higher in the gastrointestinal and genitourinary groups (21% and 13%, respectively), whereas no patients in the gynecologic or head and neck groups had weight loss >10%.
There was significant variation in treatment received by patients in this study, reflecting the diversity of primary sites and a lack of consensus regarding optimum therapy. Of the 84 patients with LD, 74 patients were considered suitable for treatment with curative intent. Fifty-three of these patients received definitive chemotherapy, 52 received definitive RT, and 29 underwent definitive surgery. Forty-three patients received definitive chemoradiotherapy (chemoRT). Of 52 patients treated with definitive RT, 48% received concurrent chemotherapy, 17.5% had sequential chemoRT, and 34.5% had trimodality therapy (surgery, RT, and chemotherapy). The RT dose ranged from 40 to 64 grays (Gy), and the most common prescription was 50 Gy in 25 fractions. Reasons for not treating with definitive RT included complete surgical resection, poor response to chemotherapy, and poor Eastern Cooperative Oncology Group (ECOG) performance status. Variation of treatment strategy by primary tumor site is illustrated in Table 3. The proportion treated with definitive RT, definitive surgery, or both, in each subsite was as follows: gynecologic (77%), genitourinary (48%), gastrointestinal (46%), head and neck (74%), and small cell carcinoma of unknown primary site (47%). In those patients who received chemotherapy, the median number of cycles was 4 for patients with LD (range, 1-6 cycles) and 3 for patients with ED (range, 1-7 cycles). The most frequently used chemotherapy regimens were cisplatin and etoposide, and carboplatin and etoposide.
|Treatment Received||Gynecology No. (%)||Genitourinary No. (%)||Gastrointestinal No. (%)||Head + Neck No. (%)||Unknown Primary Tumor No. (%)||Other Sites No. (%)||Total Cases No. (%)|
|Concurrent CRT||5 (19)||3 (27)||8 (53)||3 (21)||2 (29)||1 (100)||22 (30)|
|Sequential CRT||2 (8)||0||2 (13)||2 (14)||1 (14)||0||7 (9)|
|Surgery + CMT + RT||8 (31)||2 (18)||1 (7)||3 (21)||1 (14)||0||15 (20)|
|Surgery + RT||1 (4)||1 (9)||0||3 (21)||0||0||5 (7)|
|Surgery + CMT||6 (23)||1 (9)||2 (13)||1 (5)||1 (14)||0||11(15)|
|RT only||3 (11)||3 (27)||0||1 (5)||0||0||7 (9)|
|CMT only||0||1 (9)||1 (7)||0||1 (14)||0||3 (4)|
|Surgery only||1 (4)||0||1 (7)||1 (5)||1 (14)||0||4 (5)|
|Total||26 (100)||11 (100)||15 (100)||14 (100)||7 (100)||1 (100)||74 (100)|
The median survival rate for patients with LD was 1.4 years, with a 1-year survival rate of 63.2% (95% confidence interval [95% CI], 52.3-73%). For patients with ED, the median survival was 0.7 years, with a 1-year survival rate of 24.2% (95% CI, 12.7-41.3%). For all patients treated with RT for LD, the 1-year recurrence-free survival rate was 42.3% (95% CI, 31.2-54.3%), with a locoregional recurrence-free 1-year survival rate of 64.5% (95% CI, 51.2-75.8%) and a 1-year distant recurrence-free survival rate of 54.6% (95% CI, 42.4-66.3%). The survival rates for each subsite are presented in Table 4. Survival curves for LD and ED are shown in Figures 1 and 2, respectively.
|Disease Site||Recurrence-Free Survival at 1 Year (95% CI) Median, Years||Locoregional Recurrence-Free Survival at 1 Year (95% CI) Median, Years||Distant Recurrence-Free Survival at 1 Year (95% CI) Median, Years||Overall Survival at 1 Year (95% CI) Median, Yearsa|
|Gynecology||50% (31-69%) 1 y||70% (48-85%)b||60% (39-78%) 1.2 y||66% (48-81%) 1.4 y|
|Gastrointestinal||22% (7-50%) 0.7 y||65% (31-88%)b||29% (11-56%) 0.7 y||32% (17-52%) 0.7 y|
|Genitourinary||13% (2-54%) 0.5 y||38% (6-86%) 0.8 y||17% (3-61%) 0.6 y||44% (25-64%) 0.9 y|
|Head and neck||64% (38-84%) 1.5 y||64% (38-84%)b||93% (63-99%)b||63% (40-81%) 2.3 y|
|Unknown primary tumor||57% (23-86%) 3.4 y||69% (29-92%) 3.4 y||71% (33-93%) 3.4 y||57% (32-80%) 1.7 y|
Sites of Failure
Cumulative rates of first failure at locoregional, distant, and both distant and locoregional sites are illustrated in a competing risks analysis in Figure 3. Distant failure was more common than locoregional. The most common type of first distant recurrence was soft tissue metastasis (32%; 95% CI, 23.5-40.8%). In comparison, the brain was the site of first distant failure in only 6% of patients (95% CI, 2.4-11.6%). Of patients treated with curative intent, 49% ultimately developed soft tissue metastases (95% CI, 37-61%), compared with only 10% of patients who developed brain metastases (95% CI, 4-19%) and 16% who developed bone metastases (95% CI, 9-27%). Approximately 41% of palliatively treated patients developed bone metastases. Patients who received definitive RT were less likely to have locoregional failure (16.7% vs 35% risk of locoregional failure); however, patients treated with chemotherapy did not appear to have dramatically different rates of distant failure (Table 5).
|Site of Failure|
|Definitive Treatment||Locoregional (% Patients)||Distant (% Patients)||Locoregional + Distant (% Patients)|
Only 13% of the entire group ever developed brain metastases. The rate of brain metastases was 17% in patients with ED and, as reported above, was 10% in the definitively treated patients with LD. Only 3 of the curatively treated patients with LD (4%) developed brain metastases as the site of first distant failure.
PCI was administered to 7 patients, 6 of whom had LD. All patients treated with PCI had received platinum-based chemotherapy, and 6 patients had demonstrated a complete response in the primary site, including 4 who were restaged with PET/CT. Three patients had small cell carcinoma of unknown primary (2 axillary lymphadenopathy and 1 cervical lymphadenopathy), 2 patients had gynecologic cervical cancer, 1 had esophageal cancer, and 1 had extensive stage maxillary sinus cancer, with a complete response to chemotherapy. PCI was delivered in 2–Gy daily fractions to either 30 Gy or 36 Gy. At the time of analysis, 2 of the 7 patients had died from EPSCC. There were no brain recurrences reported.
On univariate analyses, the following prognostic risk factors for overall survival were found to be statistically significant with P <.05, (Table 6): LD, gynecologic or head and neck primary sites, ECOG performance status, weight loss, curative (definitive) versus palliative RT intent, RT dose delivered, radical chemotherapy, number of chemotherapy cycles, use of combined chemoRT, and use of PCI. On multivariate analysis, weight loss was the strongest predictor of overall survival (hazards ratio [HR], 2.131; 95%CI, 1.515-2.998 [P <.001]). LD (HR, 0.539; 95%CI, 0.311-0.936 [P = .028]), use of PCI (HR, 0.189; 95% CI, 0.046-0.783 [P = .022]), and omission of radical RT (HR, 2.465; 95% CI, 1.332-4.562 [P = .004]) were also found to be significantly associated with overall survival on multivariate analysis.
|Variable||Overall Survival N=120||Recurrence-Free Survival N=74||Overall Survival N=120||Recurrence-Free Survival N=74|
|Univariate Analysis||Multivariate Analysis|
|HR (95% CI)||P||HR (95% CI)||P||HR (95% CI)||P||HR (95% CI)||P|
|Stage||0.34 (0.22-0.53)||<.001||0.22 (0.08-0.63)||.005||0.54 (0.31-0.94)||.028|
|Gynecology site||0.49 (0.27-0.89)||.018||0.64 (0.30-1.39)||.264|
|Head and neck||0.36 (0.17-0.74)||.005||0.51 (0.21-1.23)||.134|
|Performance status||1.62 (1.19-2.21)||.002||1.19 (0.73-1.92)||.491|
|Weight loss||2.55 (1.85-3.52)||<.001||2.18 (1.38-3.44)||<.001||2.13 (1.52-2.99)||<.001||2.17 (1.38-3.41)||<.001|
|Radiotherapy||0.57 (0.37-0.88)||.01||0.50 0.27-0.90)||.021|
|Palliative vs radical RT||3.26 (1.87-5.66)||<.001||1.53 (0.36-6.44)||.563||2.47 (1.33-4.56)||.004|
|RT dose||0.98 (0.98-0.99)||<.001||0.99 (0.98-0.99)||.03|
|Chemotherapy||0.68 (0.44-1.05)||.08||0.52 0.28-0.96)||.036|
|Radical CMT||0.49 (0.30- 0.80)||.004||0.48 (0.26-0.90)||.022|
|CRT||0.56 (0.36-0.85)||.007||0.51 (0.29-0.89)||.017|
|No. of CMT cycles||0.91 (0.84-0.99)||.025||0.96 (0.88-1.04)||.319|
|PCI||0.17 (0.04-0.71)||.014||0.25 (0.06-1.01)||.052||0.19 (0.05-0.78)||.022||0.25 (0.06-1.04)||.056|
The following variables were found to be significantly associated with improved recurrence-free survival with P <.05: LD, absence of weight loss, use of RT, total RT dose, use of chemotherapy, and curative versus palliative intention of chemotherapy. PCI was not found to be significantly associated with recurrence-free survival. On multivariate analysis, weight loss was the only variable found to be statistically significantly associated with recurrence-free survival (HR, 2.167; 95% CI, 1.378-3.408 [P <.001]).
Smoking history, gender, age at diagnosis, and use or extent of surgery were not found to be significantly associated with recurrence-free or overall survival (P >.05).
To the best of our knowledge, the current study comprises the largest experience with EPSCC reported to date and has the additional advantage compared with many earlier reports that most patients were staged with CT scanning. It illustrates the often lethal nature of this disease and confirms that the predominant mode of death is due to distant metastasis. It confirms the finding tentatively reported in other series, that the rate of brain metastasis is lower than in SCLC.5 The widely different median survival rates of 1.4 years and 0.7 years, respectively, observed for LD and ED disease indicate that it is reasonable to continue using a modified version of the Veterans Affairs Lung Study Group SCLC staging system for EPSCC. The median survival for aggressively treated EPSCC patients with LD was similar to the median survival of 1.2 years for a cohort of 90 patients with SCLC treated with definitive chemoRT at our institution over a comparable time period, 63% of whom received twice daily fractionation and 51% of whom received PCI.19 However, this study does confirm that there is a significant cure rate for patients with aggressively treated EPSCC patients with LD. The 5-year survival rate was 25.4% (95% CI, 16.6-36.8%) for patients with LD and 0% for patients with ED. No disease-related deaths occurred after 4 years. Although unsatisfactory, these long-term survival results are somewhat better than reported in most series of patients with limited stage SCLC.
Median survival rates ranging from 2 months in patients with ED to 34 months in patients with LD have been reported in the literature. In a series of 101 patients described by Cicin et al, 50% of patients had LD, compared with 70% in our series.5 There was also a high proportion of carcinoma of unknown primary (31%), compared with 13% in our series. However, in that series, only 24% of patients underwent a CT scan to exclude SCLC, compared with 75% of patients in our own series.
In the current series, the strongest prognostic factor on multivariate analysis for recurrence-free and overall survival was weight loss. On multivariate analyses, the use of definitive RT was associated with improved overall survival but not recurrence-free survival, most likely because of the high incidence of distant metastases in this disease. On univariate analysis, higher RT dose and number of chemotherapy cycles were associated with improved overall survival. The addition of chemotherapy to RT was associated with improved overall and recurrence-free survival (P = .007 and P = .017, respectively). To the best of our knowledge, this finding has not been previously reported in the EPSCC literature. Surgical resection was not associated with improved outcome. In older series, a larger proportion of patients were treated with surgery, with smaller numbers receiving chemoRT. In the series reported by Haider et al,4 56% of patients underwent surgery and only 44% received chemotherapy or RT. In 1997, a study of 71 patients was published, of whom 74% were treated with surgery alone and the 5-year survival rate was only 13%.8 There are conflicting reports regarding the most appropriate treatment modality for EPSCC, with many authors recommending surgery as the primary treatment modality.13, 20, 21 Choong et al21 reported the Mayo Clinic experience of 44 patients with small cell carcinoma of the bladder treated with cystectomy, reserving chemotherapy for patients with lymph node or distant metastatic disease. In more recent literature, chemotherapy is emerging as the cornerstone of treatment.22 In a review of esophageal small cell carcinoma, the median survival for patients receiving local plus systemic treatment was 20 months compared with 5 months for those receiving local therapy only (P <.0001).23
The results of the current study indicate worse overall survival, recurrence-free survival, distant recurrence-free survival, and locoregional recurrence-free survival for patients with genitourinary and gastrointestinal primary sites. Gynecologic and head and neck tumors were associated with significantly better overall survival. However, this effect disappeared on multivariate analysis, perhaps as a result of increased use of RT in these sites. Other authors4, 12, 13 have also reported a more favorable prognosis for cervical small cell carcinoma, and Lee et al20 has reported a very poor prognosis for patients with small cell carcinoma of the gastrointestinal tract. Significant molecular differences exist between SCLC and EPSCC, which may contribute to their different clinical behavior. For example, Bcl-2 overexpression has been reported in up to 95% of SCLC compared with 33% in gastrointestinal EPSCC.17 Patterns of failure differ not only between SCLC and EPSCC, but also between various extrapulmonary tumor sites. EPSCC arising from different primary sites may need to be considered as distinct entities.
Favorable prognostic factors reported for EPSCC in other series have included ECOG performance status ≤1, and site of primary tumor.4, 9, 12, 13 In contrast to SCLC, cigarette smoking may not be a risk factor for ESPCC23 but tobacco use may be a negative prognostic factor.11 In the current series, 51% had a history of tobacco use, a proportion comparable to other series.4 This contrasts with SCLC, in which >80% of patients have a smoking history.
Suggestions for the Management of EPSCC
In the absence of prospective clinical trials, retrospective studies such as the one reported here form the only body of evidence available to help oncologists make difficult decisions concerning the treatment of rare cancers. In retrospective studies, patients are rarely managed with uniform staging and treatment strategies and, furthermore, such studies are prone to selection biases, because patients who are the fittest and who have the most LD extent are best able to tolerate more aggressive forms of therapy. There is a risk that good outcomes that are a result of good biology may be falsely attributed to more aggressive treatment. Nevertheless, EPSCC is usually a fatal disease, and for most patients, treatment intensity is simply insufficient to achieve cure. Based on our own study and a review of the literature, we suggest that it remains reasonable to manage EPSCC along broadly similar lines to SCLC. All patients should undergo a CT scan of the chest to exclude SCLC.
For EPSCC patients with ED, therapy is essentially palliative as the proportion of patients surviving for 5 years is so small. Cure is certainly possible in LD, and we recommend aggressive therapy for these patients. We do not believe that radical surgery should play a large role in the management of most patients with limited EPSCC, except in those rare cases in which a complete resection can be achieved with minimal morbidity. We suggest that chemoRT should be the primary form of management in EPSCC patients with LD. From our own experience and by analogy with SCLC, we would expect the best results with concurrent chemoRT. We suggest that the chemotherapy regimen should contain a platinum compound and etoposide, and at least 4 cycles should be given. RT should cover all known sites of primary tumor and involved lymph nodes to a dose equivalent of at least 50 Gy in 2 Gy fractions.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.