Year: 1999
Type: article
Abstract: Weinstock and Reynes1 present intriguing data from the Surveillance, Epidemiology and End Results (SEER) program about the prognosis of patients in the United States with newly diagnosed mycosis fungoides (MF). They report improved overall and relative survival over a registration period of 20 years. Improved survival was evident when recent cohorts of patients were compared with historical controls (e.g., their 1983–1992 vs. 1973–1982) and was statistically significant when date of diagnosis was considered a continuous variable (P = 0.0001). They suggest that this was not likely because of any improvements in therapy. Rather, it may have been due to earlier diagnosis, a systematic change in the case mix, a change in the biology of the disease, or the inclusion of more benign disorders in the diagnosis. The latter two suggestions are possible but speculative. In an earlier report of SEER data,2 the pathology was reviewed, and through 1984 MF was confirmed in 98% of cases. Was the pathology also reviewed for cases registered during 1985–1992? If improved overall survival were due to earlier diagnosis, two things might happen. First, the age at diagnosis might decline within some or all stage groupings during the study period. Second, patients might migrate into earlier stage groupings, being given a diagnosis prior to their developing detectable widespread skin, lymph node, blood, or visceral involvement. In the latter effect, we do not have in focus any impact that a different stage assignment would have on overall survival for those stage groupings. Instead, we are interested in changes in the proportions of patients assigned to all stage groupings. The SEER reports note that the median age of all registered cases declined slightly over the years (from 64 years to 62 years)1, 2 and that the rate of incidence of Sézary syndrome hardly changed at all (from 5% to 4%).1, 2 Unfortunately, staging information is not available for the 95–96% of patients without Sézary syndrome. Overall, then, the information is insufficient to assess whether earlier diagnosis is occurring and whether there is a lead-time bias in the calculation of overall survival. In response to the most recent report,1 we analyzed a SEER-independent data set. During 1980–1997, there were 410 patients with a new diagnosis of MF who were referred to Hamilton, where they received pathologic confirmation and, subsequently, active follow-up. This is not a population-based registry, but Hamilton is a clinical unit that accepts referrals from one-half of Canada. Population-based registries like SEER should provide more accurate statistics about rates of incidence. However, we estimate that a simple majority of patients in our apparent geographic area are referred to Hamilton. Over this 18-year period in Hamilton, there was no decline in age at diagnosis for a total of 220 referred patients with Stage IA disease, nor did the age at diagnosis decline for 97 patients with Stage IB disease or for 93 with Stage IIA–IVB. The respective correlation coefficients for these three stage groupings, when age at diagnosis was compared with calendar date of diagnosis, were +0.004 (P = 0.96 for IA), −0.15 (P = 0.26 for IB), and −0.10 (P = 0.36 for IIA–IVB). A value of the correlation coefficient near +1 or −1 indicates a strong correlation, and a value near 0 indicates no correlation. Although the population of Canada increased by about 20% over this time period, there was a modest decline in the absolute numbers of patients who presented with Stages IB–IVB (99 cases during 1980–1988 and 91 during 1989–1997). Therefore, the data set at Hamilton gives only weak support, if any, to the idea that diagnosis in Canada is being made earlier in the course of disease, across all patients or within a particular stage grouping. Furthermore, the numbers of additional patients referred to Hamilton after experiencing failure with other therapies, and who were referred for total skin electron radiation, did not change appreciably during 1980–1997. Their ages at initial diagnosis also remained constant. Therefore, the data preclude a large lead-time bias that could affect overall survival. If the improved overall survival observed in the SEER data were due to altered ascertainment, the greatest year-on-year improvements would likely correspond to important historical events and meetings and might lead to a dramatic increase in the numbers of patients with only patch and plaque disease. The SEER reports1, 2 suggest a marked increase in annual diagnoses between 1977 and 1985. One interpretation is that the true incidence of disease is rising. However, there were several international meetings and symposia, cooperative analyses of pooled data, and prominent publications during 1973–1983. The SEER data indicate that only 41 cases were registered per year during 1973–1980. In contrast, 110 cases per year were registered during 1985–1992. The data at Hamilton reveal a similar change in absolute numbers of patients. Very few patients were referred to Hamilton with a new diagnosis of MF prior to 1980. Subsequently, as we became an official working group of the Canadian Dermatology Association, annual referral became 20–30 patients, through 1994. With recent meetings, more publications, and national presentations during 1992–1998, this increased to 50–60 new patients per year. The number of patients with Stage IA disease seen during 1980–1988 was 69, but it was 151 during 1989–1997. Thus, we have seen enhanced referral of patients with Stage IA disease. A univariate Cox regression analysis with date of diagnosis coded as a continuous variable suggests a strong statistical trend (P = 0.06) in favor of more recent dates being associated with improved overall survival for all 410 patients. This is similar to what the SEER data now suggest.2 However, multivariate adjustment for age at diagnosis erodes this apparent significance (P = 0.09 for date at diagnosis). Multivariate adjustment with age at diagnosis, stage, and date of diagnosis in the equation, as 3 simultaneous independent factors, reveals only the first 2 of these to be statistically significant (both P < 0.0001, but P = 0.19 for date at diagnosis). Gender is not significant (P = 0.4). Therefore, age and stage predict for death from all causes, as one could predict from biologic principles: age predicts for death from non-MF causes, and stage predicts for death from MF. Our conclusion is that the date of diagnosis is not a fundamental prognostic factor for patients referred to Hamilton during 1980–1997. The survival analyses of the SEER data cannot be statistically adjusted for stage, and that is an important limitation. The aforementioned items might suggest that, at least in Canada, there has been no impact of therapies on overall survival. However, we have separately reported that overall survival is associated with certain treatment variables, including the administration of total skin electron radiation.3 We have also reported, in a preliminary manner, improved survival with combination therapies.4, 5 During 1980–1997 the management strategy for patients at Hamilton remained fixed. Survival differs within stage groupings between patient subgroups experiencing different therapies. However, the overall survival during 1980–1997 has remained constant because the proportions of patients experiencing various therapies has remained constant. The SEER data has not been analyzed according to therapy, and they include data from many institutions where therapy may have changed during 1973–1992. The data set at Hamilton gives little support to the idea that in the part of North America north of continental America, lead-time bias or even false diagnoses account for improved survival. We have, however, noted a marked increase in referrals of patients with Stage IA disease, perhaps in part due to increased ascertainment of same-age cases in the dermatology community. Survival was related to age at diagnosis and stage in a data set for which therapy variables were constant. In summary, the SEER data are valuable because they are population-based, but the Hamilton data highlight some of the possible interpretations of the SEER data. Glenn W. Jones M.D., M.Sc.*, Lynn D. Wilson M.D. , * Department of Radiation Oncology, Cancer Care Ontario, Hamilton, Ontario, Canada, Department of Therapeutic Radiology, Yale University School of Medicine, New Haven, Connecticut
Source: Cancer
Authors: Glenn W. Jones, Lynn D. Wilson
Institutions: Cancer Care Ontario, Yale University
Language: English
Subfield: Dermatology
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W2915809893