The Cutaneous B-Cell Lymphoma Prognostic Index: A Novel Prognostic Index Derived From a Population-Based Registry
http://www.100md.com
《临床肿瘤学》
the Department of Therapeutic Radiology, Department of Internal Medicine, Section of Medical Oncology, Yale University School of Medicine, New Haven, CT
ABSTRACT
PURPOSE: Three classification and prognostic systems exist for primary cutaneous B-cell lymphoma (PCBCL). Whereas the WHO classification defines its categories based on histology, both the European Organisation for Research and Treatment of Cancer classification and the conventional staging system consider skin site. We sought to incorporate both histology and skin site into a single prognostic index designed to predict survival in patients with PCBCL.
METHODS: The associations between histology, skin site, and survival were determined using adjusted proportional hazards analysis conducted on 926 patients with PCBCL identified in the Surveillance, Epidemiology, and End Results data, spanning 1973 to 2001.
RESULTS: Four primary CBCL prognostic index (CBCL-PI) groups were identified. Group IA included indolent histologies involving any skin site. Group IB included diffuse large B-cell histology involving favorable skin sites (head/neck, arm). Group II included diffuse large B-cell histology involving unfavorable skin sites (trunk, legs, disseminated) or immunoblastic large B-cell histology involving favorable skin sites. Group III included immunoblastic large B-cell histology involving unfavorable skin sites. The adjusted mortality hazards ratios were 1.0, 1.3 (95% CI, 0.99 to 1.7), 2.1 (95% CI, 1.6 to 2.7), and 4.5 (95% CI, 2.8 to 7.2) for groups IA/B to III, respectively. The corresponding 5-year relative survivals were 94%, 86%, 60%, and 34%.
CONCLUSION: The CBCL-PI identifies adverse combinations of histology and skin site, helping to reconcile differences in current classification and prognostic systems for PCBCL.
INTRODUCTION
Primary cutaneous B-cell lymphoma (PCBCL) is an uncommon form of extranodal non-Hodgkin's lymphoma characterized by malignant B-cells that are limited to the skin at initial diagnosis.1-10 The conventional non-Hodgkin's lymphoma staging system provides limited prognostic information for PCBCL, as patients are eligible for classification in only two categories, stages IE and IVE.11 To improve classification and prognostication of PCBCL, both the WHO12,13 and the European Organisation for Research and Treatment of Cancer (EORTC)3 have developed pathologic classification systems for PCBCL.
However, clinically meaningful differences exist between these two pathologic classifications. The WHO classification defines its categories based only on histology and adjunctive molecular studies, while the EORTC classification acknowledges leg skin site as an adverse prognostic factor. In the WHO classification, the most common category is diffuse large B-cell of any skin site,2,7 an entity which is often treated with chemotherapy and radiotherapy due to its potential for aggressive behavior.14-17 However, in the EORTC classification, diffuse large B-cell of nonleg skin sites is categorized as follicle center cell, an entity that is typically treated with radiotherapy alone due to its indolent behavior.3 Consequently, significant treatment variation may result from the conflicting classifications.
Accordingly, we sought to use population-based data to develop a valid and precise primary CBCL prognostic index (CBCL-PI). We hypothesized that differences in existing classification and prognostic systems could be reconciled by accounting for the independent contributions of both histology and skin site to patient prognosis.
METHODS
Data Source
The Surveillance, Epidemiology, and End Results (SEER) program began in 1973 and currently includes 11 population-based tumor registries and the Alaska Native Tumor Registry, representing 14% of the United States' population.18 Data on incident malignancies and follow-up is current through 2001.
Case Definition
Of 113,026 cases of non-Hodgkin's lymphoma, 6,217 were coded as originating in the skin (International Classification of Diseases –Oncology –Third Edition [ICD-O-3] C44.0 to C44.9).19 Of these, only microscopically-confirmed cases with either documented B-cell lineage (n = 908) or histology consistent with B-cell lineage (ICD-O-3 9591, 9670, 9671, 9673, 9675, 9680, 9684, 9687, 9690, 9691, 9695, 9698, 9699, 9727, 9728; n = 686) were included.19 To ensure that the cohort consisted only of PCBCL, cases with regional or distant spread were excluded (n = 251), leaving 1,343 cases for analysis. To serve as a comparison group, cases of mycosis fungoides or Sezary syndrome (ICD-O-3 9700 and 9701) were also identified (n = 3,208).19
Histology and Skin Site
Histology was coded according to the ICD-O-3,19 which is based on the WHO classification of lymphoid malignancies13 but also includes terms from older systems to permit universal coding. Immunophenotype and cytogenetic data are not reported. Indolent histologies were defined as follicular, marginal zone, small lymphocytic not otherwise specified, and lymphoplasmacytic. Diffuse large B-cell histologies were defined as both diffuse large B-cell and immunoblastic diffuse large B-cell. Skin site is coded by anatomic region (head/neck, arm, trunk, and leg), with disseminated lesions coded as either not otherwise specified or overlapping.
Covariates
Our analyses adjusted for age, sex, race (white, black, other), year of diagnosis (1973-1980, 1981 to 1990, 1991 to 2001), confirmed B-cell lineage, SEER historic stage (localized, unstaged), and treatment with radiation. Because age was exponentially associated with risk of death, both age and age-squared were included in the adjusted model. Treatment with radiation within the first 4 months of diagnosis is reported for all patients, but use of systemic chemotherapy is not reported by SEER.
Outcomes
The primary outcome was overall survival, and the secondary outcome was relative survival. Date of death is collected at the local SEER registry by matching cases to state vital statistics and, in some cases, is supplemented by data from the National Center for Health Statistics. Follow-up time was calculated from the month and year of initial diagnosis. Overall survival was calculated with the Kaplan-Meier method, and relative survival was calculated in SEER*Stat 5 (Surveillance Research Program, National Cancer Institute, http://www.seer.cancer.gov/seerstat) using race-, age-, sex-, and year of diagnosis–matched US mortality data.20 Relative survival cannot be calculated for patients with unknown race; therefore, the number of patients differed for overall survival and relative survival calculations. Follow-up was not available for 17 cases (1%).
Statistical Analysis
Age-adjusted incidence rates were calculated in SEER*Stat 5 using the year 2000 US standard population as a reference.21 Differences in rates were tested by assuming a Poisson distribution.22 Trends in incidence over time were estimated with the Joinpoint Regression Program, version 2.7 (September 2003, National Cancer Institute), with a maximum of three joinpoints.
Adjusted Cox proportional hazards models were constructed for the entire cohort and for subsets with indolent or diffuse large B-cell histology. Covariates lacking linear relationships with mortality were entered as categorical (dummy) variables, with missing values as a dummy category.
To develop the CBCL-PI, all significant pairings of histology and skin site were entered into an adjusted Cox proportional hazards model. Individual groups were ordered on the basis of mortality hazard ratios and were consolidated if groups did not offer meaningful statistical and clinical mortality differences. The likelihood ratio 2s of the initial and consolidated models were compared to ensure an adequate fit of the final model.
Differences in overall survival across CBCL-PI groups were evaluated using the log-rank test, and differences in relative survival were evaluated using the z test calculated in SEER*Stat 5.
All statistical analyses were 2-tailed with less than .05, and, unless otherwise noted, were conducted using SAS version 8.02 (SAS Inc, Cary, NC). Because our study used preexisting data without identifiers, the Yale University School of Medicine Human Investigation Committee granted an exemption from review. To protect patient identity and in concert with SEER guidelines for presentation of this public-use data set, cell counts 5 were suppressed in all text and tables.
RESULTS
Epidemiology of PCBCL
Since 1993, the age-adjusted incidence rates of PCBCL have increased, reaching a maximum of 3.9 per million (Fig 1). Comparison of age-adjusted incidence rates revealed that PCBCL was more common in men (1.4:1) and whites (2.2:1). Incidence rates steadily rose with age at diagnosis (P < .001), reaching a peak of 10.8 cases per million in those older than 80 years (Table 1).
Of the 1,343 cases identified, mean age (± standard deviation) was 64 ± 17 years, 54% of patients were male, 90% were white, and median follow-up was 3.8 years for those alive at last contact. Diffuse large B-cell was the most common histology (33%), and head or neck was the most common primary site (50%). Radiation therapy was administered in 47% (Table 2).
Histology and Skin Site As Independent Predictors of Survival
In the entire cohort, both histology (P = .0001) and skin site (P < .0001) were significant predictors of survival in adjusted analysis. Covariates including sex, race, confirmed B-cell lineage, and SEER historic stage were not significantly associated with survival.
In the subset with indolent histologies (n = 442), neither specific histologic subtype nor skin site were significant predictors of survival in adjusted analysis. However, in the subset with diffuse large B-cell histologies (n = 484), immunoblastic histologic features (hazard ratio [HR], 1.7; 95% CI, 1.1 to 2.6), trunk skin site (HR, 1.5; 95% CI, 1.02 to 2.2), leg skin site (HR, 1.8; 95% CI, 1.2 to 2.6), and disseminated skin site (HR, 1.9; 95% CI, 1.4 to 2.6) were all associated with increased mortality in adjusted analysis.
Histology and Skin Site As Composite Predictors of Survival: The CBCL-PI
In those with either indolent or diffuse large B-cell histologies (n = 926), all significant pairings of histology and skin site were entered into an adjusted model (Table 3). The model was then consolidated to form four CBCL-PI groups (Table 4). The consolidated CBCL-PI model was not inferior to the model presented in Table 3 (P = .99, where P < .05 indicates a significant difference between two models).
In the CBCL-PI model, indolent histologies involving any skin site served as the baseline, labeled as group IA. Diffuse large B-cell histology involving favorable skin sites (head/neck, arm) was associated with a marginally increased risk of death (HR, 1.3; 95% CI, 0.99 to 1.7) when compared to the baseline, and thus was assigned to group IB. Group II included diffuse large B-cell histology involving unfavorable skin sites (trunk, legs, disseminated) or immunoblastic diffuse large B-cell histology involving favorable skin sites (HR, 2.1; 95% CI, 1.6 to 2.7). Group III included immunoblastic diffuse large B-cell histology involving unfavorable skin sites (HR, 4.5; 95% CI, 2.8 to 7.2).
The mortality hazard ratios for the four CBCL-PI groups (IA, IB, II, and III, respectively) remained stable when the analysis was limited to those with confirmed B-cell lineage (n = 721; HR, 1.0, 1.5, 2.4, 5.2), localized SEER historic stage (n = 824; HR, 1.0, 1.3, 2.3, 4.9), or no history of prior malignancies (n = 865; HR, 1.0, 1.2, 2.1, 4.4).
Five-year overall survival was 81% for group IA, 72% for group IB, 48% for group II, and 27% for group III (Table 4; Fig 2). Five-year relative survival was 94% for group IA, 86% for group IB, 60% for group II, and 34% for group III (Table 4; Fig 3).
DISCUSSION
As shown in this study, the incidence of PCBCL has increased since 1993 and now approaches the incidence of mycosis fungoides, the most common cutaneous lymphoma. Previously, very little was known regarding the incidence of PCBCL in the United States. Although the Dutch Cutaneous Lymphoma Working Group found that PCBCL accounted for 19% of all cutaneous lymphomas,3 PCBCL accounted for only 4.5% of newly registered skin lymphomas in a report from three academic dermatology departments in the United States,23 and relatively few PCBCL case series have been reported by American institutions.2,16,24,25
To reconcile differences in the current classification and prognostic systems for PCBCL, we have developed a novel index, the CBCL-PI, which accounts for the independent contributions of both histology and skin site to patient outcome. Group IA, defined as indolent histologies involving any skin site, experienced a 5-year relative survival of 94%. Group IB, defined as diffuse large B-cell histology involving favorable skin sites, experienced a 5-year relative survival of 86%. Group II, defined as diffuse large B-cell histology involving unfavorable skin sites or immunoblastic large B-cell histology involving favorable skin sites, experienced a 5-year relative survival of 60%. Group III, defined as immunoblastic large B-cell histology involving unfavorable skin sites, experienced a 5-year relative survival of 34%.
When applied to PCBCL, the CBCL-PI represents a significant improvement over the conventional staging system for non-Hodgkin's lymphoma.11 Specifically, because PCBCL excludes those with systemic lymphoma, only stages IE and IVE in the conventional system apply, suggesting that the dominant prognostic factor is the distinction between localized and disseminated skin involvement. Although the CBCL-PI confirms the adverse outcome of disseminated skin site, it also emphasizes that histology and other skin sites influence outcome.
The CBCL-PI also provides important data to reconcile discrepancies between the EORTC and WHO pathologic classification systems. First, the CBCL-PI suggests that the EORTC category primary cutaneous follicle center cell lymphoma should be divided into those with follicular histology and those with diffuse large B-cell histology, as diffuse large B-cell histology was associated with a trend toward inferior survival. Second, the CBCL-PI confirms the EORTC proposition that large B-cell histology involving the leg merits a distinct prognostic category, as this skin site was associated with an inferior survival. Third, the CBCL-PI identifies other unfavorable skin sites—trunk and disseminated—that should be grouped with leg skin site due to their adverse prognoses. Finally, the CBCL-PI identifies immunoblastic diffuse large B-cell histology, a subset of "round cell histology,"6 as an independent risk factor that is worthy of incorporation into classification systems.
The CBCL-PI also provides strong validation of a recent European multi-institutional analysis of 145 patients with diffuse large B-cell histologies.6 This study identified leg skin site, multifocal skin site, and round cell histology (immunoblastic or centroblastic) as independent risk factors for poor survival.6 Other recent studies have also identified disseminated skin site26 and bcl-2 expression8 as risk factors for poor survival. Prior studies have not identified trunk skin site as a risk factor for poor survival; however, given the rarity of PCBCL, previous studies suffer from limited statistical power.6,27
Treatment options for PCBCL include radiotherapy, excision, systemic chemotherapy, rituximab, or a combination of these modalities. Given the favorable prognosis of localized disease sparing the leg, previous studies have argued that radiotherapy alone may produce reasonable results for such patients, regardless of histology.2,4,9,26 For example, in our report of 30 patients with PCBCL treated with radiotherapy alone, 5-year overall survival was 95%, and nine of 13 recurrences were limited to the skin.2 Others, however, extrapolating from experience with nodal large B-cell lymphoma, have argued that systemic chemotherapy plus involved-field radiotherapy should be utilized for those with cutaneous diffuse large B-cell lymphoma.14-17 Although not designed to address this issue definitively, our study demonstrates an inferior prognosis associated with cutaneous diffuse large B-cell lymphoma involving leg, trunk, or disseminated skin sites, thus suggesting that cytotoxic chemotherapy should at least be considered for those patients in CBCL-PI groups II and III.
While review of population-based data remains critical in efforts to enhance understanding of rare malignancies, such reports are inevitably limited by several factors, such as variation in coding practices and potential misclassification, both of which may decrease effect sizes. In addition, although our analysis adjusted for treatment with radiation, information regarding treatment with systemic chemotherapy was not available and may further modify effect sizes. Nevertheless, unique advantages of our analysis include the ability to obtain a relatively large sample size, and avoidance of the selection bias inherent in single-institution studies.
In summary, our results highlight the increasing incidence of PCBCL and identify a novel, four-tiered CBCL-PI that helps to reconcile differences in current clinical and pathologic classification systems.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Supported by the National Institutes of Health/National Institute of General Medical Sciences Medical Scientist Training Grant GM07205 (G.L.S.).
Presented at the Lymphoma Scientific Session, American Society for Therapeutic Radiology and Oncology, 46th Annual Meeting, October 6, 2004, Atlanta, GA.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Pandolfino TL, Siegel RS, Kuzel TM, et al: Primary cutaneous B-cell lymphoma: Review and current concepts. J Clin Oncol 18:2152-2168, 2000
Smith BD, Glusac EJ, McNiff JM, et al: Primary cutaneous B-cell lymphoma treated with radiotherapy: A comparison of the European Organization for Research and Treatment of Cancer and the WHO classification systems. J Clin Oncol 22:634-639, 2004
Willemze R, Kerl H, Sterry W, et al: EORTC classification for primary cutaneous lymphomas: A proposal from the Cutaneous Lymphoma Study Group of the European Organization for Research and Treatment of Cancer. Blood 90:354-371, 1997
Rijlaarsdam JU, Toonstra J, Meijer OW, et al: Treatment of primary cutaneous B-cell lymphomas of follicle center cell origin: A clinical follow-up study of 55 patients treated with radiotherapy or polychemotherapy. J Clin Oncol 14:549-555, 1996
Piccinno R, Caccialanza M, Berti E: Dermatologic radiotherapy of primary cutaneous follicle center cell lymphoma. Eur J Dermatol 13:49-52, 2003
Grange F, Bekkenk MW, Wechsler J, et al: Prognostic factors in primary cutaneous large B-cell lymphomas: A European multicenter study. J Clin Oncol 19:3602-3610, 2001
Fink-Puches R, Zenahlik P, Back B, et al: Primary cutaneous lymphomas: Applicability of current classification schemes (European Organization for Research and Treatment of Cancer, World Health Organization) based on clinicopathologic features observed in a large group of patients. Blood 99:800-805, 2002
Grange F, Petrella T, Beylot-Barry M, et al: Bcl-2 protein expression is the strongest independent prognostic factor of survival in primary cutaneous large B-cell lymphomas. Blood 103:3662-3668, 2004
Bekkenk MW, Vermeer MH, Geerts ML, et al: Treatment of multifocal primary cutaneous B-cell lymphoma: A clinical follow-up study of 29 patients. J Clin Oncol 17:2471-2478, 1999
Santucci M, Pimpinelli N, Arganini L: Primary cutaneous B-cell lymphoma: A unique type of low-grade lymphoma—Clinicopathologic and immunologic study of 83 cases. Cancer 67:2311-2326, 1991
Greene FL, Page DL, Fleming ID, et al: AJCC Cancer Staging Handbook, Sixth Edition. New York, NY, Springer, 2002
Sander CA, Flaig MJ, Jaffe ES: Cutaneous manifestations of lymphoma: A clinical guide based on the WHO classification. Clin Lymphoma 2:86-102, 2001
Harris NL, Jaffe ES, Diebold J, et al: The World Health Organization classification of neoplastic diseases of the haematopoietic and lymphoid tissues: Report of the Clinical Advisory Committee Meeting, Airlie House, Virginia, November 1997. Histopathology 36:69-86, 2000
Horning SJ, Weller E, Kim K, et al: Chemotherapy with or without radiotherapy in limited-stage diffuse aggressive non-Hodgkin's lymphoma: Eastern Cooperative Oncology Group study 1484. J Clin Oncol 22:3032-3038, 2004
Miller TP, Dahlberg S, Cassady JR, et al: Chemotherapy alone compared with chemotherapy plus radiotherapy for localized intermediate- and high-grade non-Hodgkin's lymphoma. N Engl J Med 339:21-26, 1998
Sarris AH, Braunschweig I, Medeiros LJ, et al: Primary cutaneous non-Hodgkin's lymphoma of Ann Arbor stage I: Preferential cutaneous relapses but high cure rate with doxorubicin-based therapy. J Clin Oncol 19:398-405, 2001
Connors JM, Hsi ED, Foss FM: Lymphoma of the skin. Hematology (Am Soc Hematol Educ Program) 263-282, 2002
Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Incidence: SEER 11 Regs + AK Public-Use, Nov 2003 Sub (1973-2001 varying), National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2004, based on the November 2003 submission.
Fritz A, Percy C, Jack A, et al: International Classification of Diseases for Oncology (ed 3). Geneva, Switzerland, World Health Organization, 2000
Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Mortality: All COD, Public-Use With State, Total U.S. (1969-2001), National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2004
Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Populations: Total U.S. (1969-2001), National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2004
National Center for Health Statistics: Births, marriages, divorces, and deaths for January 1991. Monthly Vital Statistics Report, vol 40, No. 1, Hyattsville, MD, Public Health Service, 1991
Zackheim HS, Vonderheid EC, Ramsay DL, et al: Relative frequency of various forms of primary cutaneous lymphomas. J Am Acad Dermatol 43:793-796, 2000
Mirza I, Macpherson N, Paproski S, et al: Primary cutaneous follicular lymphoma: An assessment of clinical, histopathologic, immunophenotypic, and molecular features. J Clin Oncol 20:647-655, 2002
Aguilera NS, Tomaszewski MM, Moad JC, et al: Cutaneous follicle center lymphoma: A clinicopathologic study of 19 cases. Mod Pathol 14:828-835, 2001
Eich HT, Eich D, Micke O, et al: Long-term efficacy, curative potential, and prognostic factors of radiotherapy in primary cutaneous B-cell lymphoma. Int J Radiat Oncol Biol Phys 55:899-906, 2003
Berti E, Alessi E, Caputo R, et al: Reticulohistiocytoma of the dorsum. J Am Acad Dermatol 19:259-272, 1988(Benjamin D. Smith, Grace )
ABSTRACT
PURPOSE: Three classification and prognostic systems exist for primary cutaneous B-cell lymphoma (PCBCL). Whereas the WHO classification defines its categories based on histology, both the European Organisation for Research and Treatment of Cancer classification and the conventional staging system consider skin site. We sought to incorporate both histology and skin site into a single prognostic index designed to predict survival in patients with PCBCL.
METHODS: The associations between histology, skin site, and survival were determined using adjusted proportional hazards analysis conducted on 926 patients with PCBCL identified in the Surveillance, Epidemiology, and End Results data, spanning 1973 to 2001.
RESULTS: Four primary CBCL prognostic index (CBCL-PI) groups were identified. Group IA included indolent histologies involving any skin site. Group IB included diffuse large B-cell histology involving favorable skin sites (head/neck, arm). Group II included diffuse large B-cell histology involving unfavorable skin sites (trunk, legs, disseminated) or immunoblastic large B-cell histology involving favorable skin sites. Group III included immunoblastic large B-cell histology involving unfavorable skin sites. The adjusted mortality hazards ratios were 1.0, 1.3 (95% CI, 0.99 to 1.7), 2.1 (95% CI, 1.6 to 2.7), and 4.5 (95% CI, 2.8 to 7.2) for groups IA/B to III, respectively. The corresponding 5-year relative survivals were 94%, 86%, 60%, and 34%.
CONCLUSION: The CBCL-PI identifies adverse combinations of histology and skin site, helping to reconcile differences in current classification and prognostic systems for PCBCL.
INTRODUCTION
Primary cutaneous B-cell lymphoma (PCBCL) is an uncommon form of extranodal non-Hodgkin's lymphoma characterized by malignant B-cells that are limited to the skin at initial diagnosis.1-10 The conventional non-Hodgkin's lymphoma staging system provides limited prognostic information for PCBCL, as patients are eligible for classification in only two categories, stages IE and IVE.11 To improve classification and prognostication of PCBCL, both the WHO12,13 and the European Organisation for Research and Treatment of Cancer (EORTC)3 have developed pathologic classification systems for PCBCL.
However, clinically meaningful differences exist between these two pathologic classifications. The WHO classification defines its categories based only on histology and adjunctive molecular studies, while the EORTC classification acknowledges leg skin site as an adverse prognostic factor. In the WHO classification, the most common category is diffuse large B-cell of any skin site,2,7 an entity which is often treated with chemotherapy and radiotherapy due to its potential for aggressive behavior.14-17 However, in the EORTC classification, diffuse large B-cell of nonleg skin sites is categorized as follicle center cell, an entity that is typically treated with radiotherapy alone due to its indolent behavior.3 Consequently, significant treatment variation may result from the conflicting classifications.
Accordingly, we sought to use population-based data to develop a valid and precise primary CBCL prognostic index (CBCL-PI). We hypothesized that differences in existing classification and prognostic systems could be reconciled by accounting for the independent contributions of both histology and skin site to patient prognosis.
METHODS
Data Source
The Surveillance, Epidemiology, and End Results (SEER) program began in 1973 and currently includes 11 population-based tumor registries and the Alaska Native Tumor Registry, representing 14% of the United States' population.18 Data on incident malignancies and follow-up is current through 2001.
Case Definition
Of 113,026 cases of non-Hodgkin's lymphoma, 6,217 were coded as originating in the skin (International Classification of Diseases –Oncology –Third Edition [ICD-O-3] C44.0 to C44.9).19 Of these, only microscopically-confirmed cases with either documented B-cell lineage (n = 908) or histology consistent with B-cell lineage (ICD-O-3 9591, 9670, 9671, 9673, 9675, 9680, 9684, 9687, 9690, 9691, 9695, 9698, 9699, 9727, 9728; n = 686) were included.19 To ensure that the cohort consisted only of PCBCL, cases with regional or distant spread were excluded (n = 251), leaving 1,343 cases for analysis. To serve as a comparison group, cases of mycosis fungoides or Sezary syndrome (ICD-O-3 9700 and 9701) were also identified (n = 3,208).19
Histology and Skin Site
Histology was coded according to the ICD-O-3,19 which is based on the WHO classification of lymphoid malignancies13 but also includes terms from older systems to permit universal coding. Immunophenotype and cytogenetic data are not reported. Indolent histologies were defined as follicular, marginal zone, small lymphocytic not otherwise specified, and lymphoplasmacytic. Diffuse large B-cell histologies were defined as both diffuse large B-cell and immunoblastic diffuse large B-cell. Skin site is coded by anatomic region (head/neck, arm, trunk, and leg), with disseminated lesions coded as either not otherwise specified or overlapping.
Covariates
Our analyses adjusted for age, sex, race (white, black, other), year of diagnosis (1973-1980, 1981 to 1990, 1991 to 2001), confirmed B-cell lineage, SEER historic stage (localized, unstaged), and treatment with radiation. Because age was exponentially associated with risk of death, both age and age-squared were included in the adjusted model. Treatment with radiation within the first 4 months of diagnosis is reported for all patients, but use of systemic chemotherapy is not reported by SEER.
Outcomes
The primary outcome was overall survival, and the secondary outcome was relative survival. Date of death is collected at the local SEER registry by matching cases to state vital statistics and, in some cases, is supplemented by data from the National Center for Health Statistics. Follow-up time was calculated from the month and year of initial diagnosis. Overall survival was calculated with the Kaplan-Meier method, and relative survival was calculated in SEER*Stat 5 (Surveillance Research Program, National Cancer Institute, http://www.seer.cancer.gov/seerstat) using race-, age-, sex-, and year of diagnosis–matched US mortality data.20 Relative survival cannot be calculated for patients with unknown race; therefore, the number of patients differed for overall survival and relative survival calculations. Follow-up was not available for 17 cases (1%).
Statistical Analysis
Age-adjusted incidence rates were calculated in SEER*Stat 5 using the year 2000 US standard population as a reference.21 Differences in rates were tested by assuming a Poisson distribution.22 Trends in incidence over time were estimated with the Joinpoint Regression Program, version 2.7 (September 2003, National Cancer Institute), with a maximum of three joinpoints.
Adjusted Cox proportional hazards models were constructed for the entire cohort and for subsets with indolent or diffuse large B-cell histology. Covariates lacking linear relationships with mortality were entered as categorical (dummy) variables, with missing values as a dummy category.
To develop the CBCL-PI, all significant pairings of histology and skin site were entered into an adjusted Cox proportional hazards model. Individual groups were ordered on the basis of mortality hazard ratios and were consolidated if groups did not offer meaningful statistical and clinical mortality differences. The likelihood ratio 2s of the initial and consolidated models were compared to ensure an adequate fit of the final model.
Differences in overall survival across CBCL-PI groups were evaluated using the log-rank test, and differences in relative survival were evaluated using the z test calculated in SEER*Stat 5.
All statistical analyses were 2-tailed with less than .05, and, unless otherwise noted, were conducted using SAS version 8.02 (SAS Inc, Cary, NC). Because our study used preexisting data without identifiers, the Yale University School of Medicine Human Investigation Committee granted an exemption from review. To protect patient identity and in concert with SEER guidelines for presentation of this public-use data set, cell counts 5 were suppressed in all text and tables.
RESULTS
Epidemiology of PCBCL
Since 1993, the age-adjusted incidence rates of PCBCL have increased, reaching a maximum of 3.9 per million (Fig 1). Comparison of age-adjusted incidence rates revealed that PCBCL was more common in men (1.4:1) and whites (2.2:1). Incidence rates steadily rose with age at diagnosis (P < .001), reaching a peak of 10.8 cases per million in those older than 80 years (Table 1).
Of the 1,343 cases identified, mean age (± standard deviation) was 64 ± 17 years, 54% of patients were male, 90% were white, and median follow-up was 3.8 years for those alive at last contact. Diffuse large B-cell was the most common histology (33%), and head or neck was the most common primary site (50%). Radiation therapy was administered in 47% (Table 2).
Histology and Skin Site As Independent Predictors of Survival
In the entire cohort, both histology (P = .0001) and skin site (P < .0001) were significant predictors of survival in adjusted analysis. Covariates including sex, race, confirmed B-cell lineage, and SEER historic stage were not significantly associated with survival.
In the subset with indolent histologies (n = 442), neither specific histologic subtype nor skin site were significant predictors of survival in adjusted analysis. However, in the subset with diffuse large B-cell histologies (n = 484), immunoblastic histologic features (hazard ratio [HR], 1.7; 95% CI, 1.1 to 2.6), trunk skin site (HR, 1.5; 95% CI, 1.02 to 2.2), leg skin site (HR, 1.8; 95% CI, 1.2 to 2.6), and disseminated skin site (HR, 1.9; 95% CI, 1.4 to 2.6) were all associated with increased mortality in adjusted analysis.
Histology and Skin Site As Composite Predictors of Survival: The CBCL-PI
In those with either indolent or diffuse large B-cell histologies (n = 926), all significant pairings of histology and skin site were entered into an adjusted model (Table 3). The model was then consolidated to form four CBCL-PI groups (Table 4). The consolidated CBCL-PI model was not inferior to the model presented in Table 3 (P = .99, where P < .05 indicates a significant difference between two models).
In the CBCL-PI model, indolent histologies involving any skin site served as the baseline, labeled as group IA. Diffuse large B-cell histology involving favorable skin sites (head/neck, arm) was associated with a marginally increased risk of death (HR, 1.3; 95% CI, 0.99 to 1.7) when compared to the baseline, and thus was assigned to group IB. Group II included diffuse large B-cell histology involving unfavorable skin sites (trunk, legs, disseminated) or immunoblastic diffuse large B-cell histology involving favorable skin sites (HR, 2.1; 95% CI, 1.6 to 2.7). Group III included immunoblastic diffuse large B-cell histology involving unfavorable skin sites (HR, 4.5; 95% CI, 2.8 to 7.2).
The mortality hazard ratios for the four CBCL-PI groups (IA, IB, II, and III, respectively) remained stable when the analysis was limited to those with confirmed B-cell lineage (n = 721; HR, 1.0, 1.5, 2.4, 5.2), localized SEER historic stage (n = 824; HR, 1.0, 1.3, 2.3, 4.9), or no history of prior malignancies (n = 865; HR, 1.0, 1.2, 2.1, 4.4).
Five-year overall survival was 81% for group IA, 72% for group IB, 48% for group II, and 27% for group III (Table 4; Fig 2). Five-year relative survival was 94% for group IA, 86% for group IB, 60% for group II, and 34% for group III (Table 4; Fig 3).
DISCUSSION
As shown in this study, the incidence of PCBCL has increased since 1993 and now approaches the incidence of mycosis fungoides, the most common cutaneous lymphoma. Previously, very little was known regarding the incidence of PCBCL in the United States. Although the Dutch Cutaneous Lymphoma Working Group found that PCBCL accounted for 19% of all cutaneous lymphomas,3 PCBCL accounted for only 4.5% of newly registered skin lymphomas in a report from three academic dermatology departments in the United States,23 and relatively few PCBCL case series have been reported by American institutions.2,16,24,25
To reconcile differences in the current classification and prognostic systems for PCBCL, we have developed a novel index, the CBCL-PI, which accounts for the independent contributions of both histology and skin site to patient outcome. Group IA, defined as indolent histologies involving any skin site, experienced a 5-year relative survival of 94%. Group IB, defined as diffuse large B-cell histology involving favorable skin sites, experienced a 5-year relative survival of 86%. Group II, defined as diffuse large B-cell histology involving unfavorable skin sites or immunoblastic large B-cell histology involving favorable skin sites, experienced a 5-year relative survival of 60%. Group III, defined as immunoblastic large B-cell histology involving unfavorable skin sites, experienced a 5-year relative survival of 34%.
When applied to PCBCL, the CBCL-PI represents a significant improvement over the conventional staging system for non-Hodgkin's lymphoma.11 Specifically, because PCBCL excludes those with systemic lymphoma, only stages IE and IVE in the conventional system apply, suggesting that the dominant prognostic factor is the distinction between localized and disseminated skin involvement. Although the CBCL-PI confirms the adverse outcome of disseminated skin site, it also emphasizes that histology and other skin sites influence outcome.
The CBCL-PI also provides important data to reconcile discrepancies between the EORTC and WHO pathologic classification systems. First, the CBCL-PI suggests that the EORTC category primary cutaneous follicle center cell lymphoma should be divided into those with follicular histology and those with diffuse large B-cell histology, as diffuse large B-cell histology was associated with a trend toward inferior survival. Second, the CBCL-PI confirms the EORTC proposition that large B-cell histology involving the leg merits a distinct prognostic category, as this skin site was associated with an inferior survival. Third, the CBCL-PI identifies other unfavorable skin sites—trunk and disseminated—that should be grouped with leg skin site due to their adverse prognoses. Finally, the CBCL-PI identifies immunoblastic diffuse large B-cell histology, a subset of "round cell histology,"6 as an independent risk factor that is worthy of incorporation into classification systems.
The CBCL-PI also provides strong validation of a recent European multi-institutional analysis of 145 patients with diffuse large B-cell histologies.6 This study identified leg skin site, multifocal skin site, and round cell histology (immunoblastic or centroblastic) as independent risk factors for poor survival.6 Other recent studies have also identified disseminated skin site26 and bcl-2 expression8 as risk factors for poor survival. Prior studies have not identified trunk skin site as a risk factor for poor survival; however, given the rarity of PCBCL, previous studies suffer from limited statistical power.6,27
Treatment options for PCBCL include radiotherapy, excision, systemic chemotherapy, rituximab, or a combination of these modalities. Given the favorable prognosis of localized disease sparing the leg, previous studies have argued that radiotherapy alone may produce reasonable results for such patients, regardless of histology.2,4,9,26 For example, in our report of 30 patients with PCBCL treated with radiotherapy alone, 5-year overall survival was 95%, and nine of 13 recurrences were limited to the skin.2 Others, however, extrapolating from experience with nodal large B-cell lymphoma, have argued that systemic chemotherapy plus involved-field radiotherapy should be utilized for those with cutaneous diffuse large B-cell lymphoma.14-17 Although not designed to address this issue definitively, our study demonstrates an inferior prognosis associated with cutaneous diffuse large B-cell lymphoma involving leg, trunk, or disseminated skin sites, thus suggesting that cytotoxic chemotherapy should at least be considered for those patients in CBCL-PI groups II and III.
While review of population-based data remains critical in efforts to enhance understanding of rare malignancies, such reports are inevitably limited by several factors, such as variation in coding practices and potential misclassification, both of which may decrease effect sizes. In addition, although our analysis adjusted for treatment with radiation, information regarding treatment with systemic chemotherapy was not available and may further modify effect sizes. Nevertheless, unique advantages of our analysis include the ability to obtain a relatively large sample size, and avoidance of the selection bias inherent in single-institution studies.
In summary, our results highlight the increasing incidence of PCBCL and identify a novel, four-tiered CBCL-PI that helps to reconcile differences in current clinical and pathologic classification systems.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Supported by the National Institutes of Health/National Institute of General Medical Sciences Medical Scientist Training Grant GM07205 (G.L.S.).
Presented at the Lymphoma Scientific Session, American Society for Therapeutic Radiology and Oncology, 46th Annual Meeting, October 6, 2004, Atlanta, GA.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Pandolfino TL, Siegel RS, Kuzel TM, et al: Primary cutaneous B-cell lymphoma: Review and current concepts. J Clin Oncol 18:2152-2168, 2000
Smith BD, Glusac EJ, McNiff JM, et al: Primary cutaneous B-cell lymphoma treated with radiotherapy: A comparison of the European Organization for Research and Treatment of Cancer and the WHO classification systems. J Clin Oncol 22:634-639, 2004
Willemze R, Kerl H, Sterry W, et al: EORTC classification for primary cutaneous lymphomas: A proposal from the Cutaneous Lymphoma Study Group of the European Organization for Research and Treatment of Cancer. Blood 90:354-371, 1997
Rijlaarsdam JU, Toonstra J, Meijer OW, et al: Treatment of primary cutaneous B-cell lymphomas of follicle center cell origin: A clinical follow-up study of 55 patients treated with radiotherapy or polychemotherapy. J Clin Oncol 14:549-555, 1996
Piccinno R, Caccialanza M, Berti E: Dermatologic radiotherapy of primary cutaneous follicle center cell lymphoma. Eur J Dermatol 13:49-52, 2003
Grange F, Bekkenk MW, Wechsler J, et al: Prognostic factors in primary cutaneous large B-cell lymphomas: A European multicenter study. J Clin Oncol 19:3602-3610, 2001
Fink-Puches R, Zenahlik P, Back B, et al: Primary cutaneous lymphomas: Applicability of current classification schemes (European Organization for Research and Treatment of Cancer, World Health Organization) based on clinicopathologic features observed in a large group of patients. Blood 99:800-805, 2002
Grange F, Petrella T, Beylot-Barry M, et al: Bcl-2 protein expression is the strongest independent prognostic factor of survival in primary cutaneous large B-cell lymphomas. Blood 103:3662-3668, 2004
Bekkenk MW, Vermeer MH, Geerts ML, et al: Treatment of multifocal primary cutaneous B-cell lymphoma: A clinical follow-up study of 29 patients. J Clin Oncol 17:2471-2478, 1999
Santucci M, Pimpinelli N, Arganini L: Primary cutaneous B-cell lymphoma: A unique type of low-grade lymphoma—Clinicopathologic and immunologic study of 83 cases. Cancer 67:2311-2326, 1991
Greene FL, Page DL, Fleming ID, et al: AJCC Cancer Staging Handbook, Sixth Edition. New York, NY, Springer, 2002
Sander CA, Flaig MJ, Jaffe ES: Cutaneous manifestations of lymphoma: A clinical guide based on the WHO classification. Clin Lymphoma 2:86-102, 2001
Harris NL, Jaffe ES, Diebold J, et al: The World Health Organization classification of neoplastic diseases of the haematopoietic and lymphoid tissues: Report of the Clinical Advisory Committee Meeting, Airlie House, Virginia, November 1997. Histopathology 36:69-86, 2000
Horning SJ, Weller E, Kim K, et al: Chemotherapy with or without radiotherapy in limited-stage diffuse aggressive non-Hodgkin's lymphoma: Eastern Cooperative Oncology Group study 1484. J Clin Oncol 22:3032-3038, 2004
Miller TP, Dahlberg S, Cassady JR, et al: Chemotherapy alone compared with chemotherapy plus radiotherapy for localized intermediate- and high-grade non-Hodgkin's lymphoma. N Engl J Med 339:21-26, 1998
Sarris AH, Braunschweig I, Medeiros LJ, et al: Primary cutaneous non-Hodgkin's lymphoma of Ann Arbor stage I: Preferential cutaneous relapses but high cure rate with doxorubicin-based therapy. J Clin Oncol 19:398-405, 2001
Connors JM, Hsi ED, Foss FM: Lymphoma of the skin. Hematology (Am Soc Hematol Educ Program) 263-282, 2002
Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Incidence: SEER 11 Regs + AK Public-Use, Nov 2003 Sub (1973-2001 varying), National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2004, based on the November 2003 submission.
Fritz A, Percy C, Jack A, et al: International Classification of Diseases for Oncology (ed 3). Geneva, Switzerland, World Health Organization, 2000
Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Mortality: All COD, Public-Use With State, Total U.S. (1969-2001), National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2004
Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Populations: Total U.S. (1969-2001), National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2004
National Center for Health Statistics: Births, marriages, divorces, and deaths for January 1991. Monthly Vital Statistics Report, vol 40, No. 1, Hyattsville, MD, Public Health Service, 1991
Zackheim HS, Vonderheid EC, Ramsay DL, et al: Relative frequency of various forms of primary cutaneous lymphomas. J Am Acad Dermatol 43:793-796, 2000
Mirza I, Macpherson N, Paproski S, et al: Primary cutaneous follicular lymphoma: An assessment of clinical, histopathologic, immunophenotypic, and molecular features. J Clin Oncol 20:647-655, 2002
Aguilera NS, Tomaszewski MM, Moad JC, et al: Cutaneous follicle center lymphoma: A clinicopathologic study of 19 cases. Mod Pathol 14:828-835, 2001
Eich HT, Eich D, Micke O, et al: Long-term efficacy, curative potential, and prognostic factors of radiotherapy in primary cutaneous B-cell lymphoma. Int J Radiat Oncol Biol Phys 55:899-906, 2003
Berti E, Alessi E, Caputo R, et al: Reticulohistiocytoma of the dorsum. J Am Acad Dermatol 19:259-272, 1988(Benjamin D. Smith, Grace )