Fax: (011) 612 9550 6316
Prognosis for patients with thin cutaneous melanoma
Long-term survival data from the New South Wales Central Cancer Registry and the Sydney Melanoma Unit
Article first published online: 20 AUG 2003
Copyright © 2003 American Cancer Society
Volume 98, Issue 6, pages 1223–1231, 15 September 2003
How to Cite
McKinnon, J. G., Yu, X. Q., McCarthy, W. H. and Thompson, J. F. (2003), Prognosis for patients with thin cutaneous melanoma. Cancer, 98: 1223–1231. doi: 10.1002/cncr.11624
- Issue published online: 5 SEP 2003
- Article first published online: 20 AUG 2003
- Manuscript Accepted: 6 JUN 2003
- Manuscript Revised: 4 JUN 2003
- Manuscript Received: 5 FEB 2003
- Melanoma Foundation of the University of Sydney
- Walter C. MacKenzie-Scotiabank Fellowship
- tumor thickness
Estimates of long-term survival for patients with thin (≤ 1 mm) primary cutaneous melanomas vary widely. Two separate methods were used to study the survival of patients with melanoma from New South Wales (NSW), Australia, and from the Sydney Melanoma Unit (SMU).
The NSW Central Cancer Registry (NSWCCR) provided data on all patients who were diagnosed with cutaneous melanomas that measured ≤ 1 mm thick between 1983 and 1998, inclusive. Patients with metastases at the time of diagnosis were not included, leaving 18,088 patients for analysis. The SMU data base was analyzed to extract data for all patients with thin melanomas who met the same criteria from 1979 to 1998, inclusive. All patients who had their primary tumors treated definitively elsewhere were excluded, leaving 2746 patients for analysis. Ten-year Kaplan–Meier survival rates were calculated, and significant differences were determined using log-rank analysis. Prognostic factors were evaluated with Cox proportional hazards analysis.
The NSWCCR analysis revealed a 10-year survival rate of 96.4%. The 10-year survival rate for patients at SMU was 92.7%. Among the patients at SMU who died, the median time to recurrence was 49.8 months, and the median time to death was 65.9 months. The 10-year survival for patients at SMU who had lesions that measured ≤ 0.75 mm was 96.9% compared with 84.3% for patients who had lesions that measured 0.76–1.0 mm. For patients who had ulcerated melanomas measuring ≤ 1 mm thick, the 10-year survival rate was 83%, compared with 92.3% for patients who had nonulcerated melanomas.
The results of the current study confirmed the excellent survival rate for patients with thin melanomas. Higher-risk subsets of patients who may warrant consideration for aggressive investigation and treatment are identifiable. Cancer 2003;98:1223–31. © 2003 American Cancer Society.
Many attempts have been made to determine prognosis for patients with thin primary cutaneous melanomas. The largest study to date analyzed data from 17,600 patients from 13 different cancer centers and cooperative groups to formulate a new American Joint Committee on Cancer (AJCC) staging system.1, 2 This system, which also was adopted by the International Union Against Cancer (UICC), will be used worldwide to stage and guide treatment for virtually all patients with this disease. In the new staging system, tumors that measure ≤ 1 mm thick are classified as T1a if they are nonulcerated or are Level II or III lesions and T1b if they are ulcerated or are Level IV or V lesions. The reported actuarial survival rates at 10 years for these groups were 87.9% (n = 4510) and 83.1% (n = 1380), respectively.
Although they were documented exhaustively and were based on a very large sample of patients (approximately half from the Sydney Melanoma Unit [SMU]), the survival rates predicted by analysis for the new staging system appear to be markedly poorer compared with the survival rates reported in other large series. For example, Buttner et al. published survival statistics for patients from four university centers in Germany.3 In that report, the 10-year survival rate for patients with melanomas measuring ≤ 1 mm thick was 94.5% (n = 1849). These data were updated in 2002, with similar results.4 Ulceration had no impact on survival in patients with thin melanomas. To our knowledge, there are scant published data from population-based registries. The few studies that exist are difficult to compare because of low numbers, short follow-up, and different cut-off points for survival.5–10 However data reported from the South Australian Cancer Registry also suggest that the long-term survival of patients with thin melanomas is much better than predicted by the AJCC data.11 The 10-year survival rate for patients with lesions that measured ≤ 0.50 mm reportedly was 98.0% (n = 2737); for patients with lesions that measured 0.51–0.75 mm, it was 96.6% (n = 1400); and, for patients with lesions that measured 0.76–1.0 mm, it was 91.5% (n = 898).
It is important for both patients with melanoma and their treating physicians to have a realistic estimate of long-term survival. Although it has been shown clearly that there are no natural biologic break points for thickness, it is convenient to use ≤ 1 mm as a cut-off point to define early lesions.2, 3, 12, 13 It is congruent with the AJCC staging system and it is a useful guide for determining both margin of primary tumor excision and eligibility for sentinel lymph node biopsy or adjuvant therapy.14–16 It is self-evident that patients with melanoma are entitled to an estimate of cure rate that is easy to understand and that accurately reflects their prognosis.
To obtain as accurate an estimation of survival as possible for patients with thin melanomas, we extracted data both from the New South Wales Central Cancer Registry (NSWCCR) and from the SMU data base. The latter records information collected prospectively for large numbers of patients with melanoma who are followed for long periods.
MATERIALS AND METHODS
In the State of New South Wales, Australia, the law requires that all patients with cancer must be reported to the NSWCCR. The NSWCCR was requested to provide data on all patients with melanomas who had the following characteristics: 1) diagnosis between the years 1983 and 1998, inclusive; 2) primary, invasive, cutaneous melanoma measuring ≤ 1 mm in thickness; 3) no multiple primary melanomas or metastatic disease at the time of diagnosis; 4) up-to-date follow-up, even if the patient had left the State of New South Wales but still lived in Australia; 5) restaged melanoma, if necessary, on the basis of any additional pathologic information subsequently supplied by review at a tertiary melanoma treatment center; and 6) diagnosis that was not made exclusively by death certificate. Only patients who were entered into the registry at the time of pathologic diagnosis were included in the analysis.
Data for patients who were treated prior to 1983 were judged likely to be unreliable and thus were not sought. A total of 34,107 potentially eligible patients with melanoma were identified from the registry. Details are shown in Table 1. Of these 34,107 patients, 18,088 met the criteria described above and were included in the analysis. On the basis of these data, the registry then supplied melanoma-specific 10-year survival rates calculated according to the Kaplan–Meier method. The NSWCCR does not actively follow patients with melanoma. Survival is determined by matching against death records from the State Registrar of Births, Deaths, and Marriages and from the National Death Index. All living patients are considered censored data points. Median follow-up data, therefore, are not available.
|Thickness group||Total||Failed||No. of censored patients (%)|
|No information on thickness||5647||1750||3897 (69.01)|
|Thickness ≤ 1 mm and localized||18,088||471||17,617 (97.40)|
|Thickness > 1 mm||10,372||2049||8323 (80.24)|
SMU data were extracted from the prospectively collected information in its database to include all patients who had the following characteristics: 1) diagnosis over a 20 year period from 1979 to 1998, inclusive, thus allowing adequate follow-up; 2) localized, invasive, cutaneous melanoma that measured ≤ 1 mm in thickness; 3) did not have multiple primary melanomas on diagnosis or on subsequent follow-up; 4) posttreatment follow-up; and 5) primary disease that was treated definitively at the SMU. All patients who had their primary disease treated elsewhere were excluded to avoid a referral bias (e.g., when they were referred after they developed metastases). However, patients who only underwent a diagnostic biopsy elsewhere were included. A total of 2746 SMU patients were available for analysis.
All deaths that occurred among SMU patients were audited by chart review. Differences between the NSWCCR population and the SMU group were tested using the Wilcoxon rank-sum test. Melanoma-specific 10-year survival rates were calculated according to the Kaplan–Meier method. Differences between survival rates were tested with log-rank analysis. Predictors of survival were tested with Cox proportional hazards analysis. Thickness, age, and mitotic rate (mitoses per mm2) were analyzed as continuous variables.
In the NSWCCR population, there were 9656 males and 8432 females; their mean and median ages were 53 years and 54 years, respectively. The mean thickness of the melanomas was 0.55 mm, with a median of 0.50 mm. Melanoma-specific survival data from the NSWCCR are presented in Table 2 and are shown in Figure 1. For the group with localized melanomas measuring ≤ 1 mm in thickness, the 10-year survival rate was 96.4% (95% confidence interval [CI], 96.0–96.7%), and the 15-year survival rate was 95.0% (95% CI, 94.4–96.5%). For comparison, survival also is shown for patients who had tumors measuring > 1 mm in thickness as well as the group for whom microstaging information was not available. Survival according to further breakdown of thickness groups also is shown in Figure 1. Other prognostic factors, such as ulceration, were not available from the NSWCCR.
|Thickness (mm)||No. of patients||10 yr survival (%)||15 yr survival (%)|
There were 1343 males and 1404 females in the group of patients from the SMU; their mean and median ages were 47.1 years and 45.9 years, respectively. The mean and median thickness of melanomas was 0.61 mm and 0.60 mm, respectively. Both age and thickness differed significantly from the NSWCCR patient group (P ≤ 0.0001). The median follow-up for the entire group was 49.1 months (range, 1–269 months). Overall melanoma-specific survival (± standard error [SE]) was 92.7% ± 1.8% at 10 years and 89.3% ± 3.0% at 15 years. Data are shown in Table 3. A Kaplan–Meier survival curve is shown in Figure 2.
|No. of patients||2746|
|10 yr survival rate (%)||92.7|
|Standard error (%)||1.77|
The 10-year survival rates for patient groups (± SE) with and without ulceration were 84% ± 9.5% and 92.3% ± 2.1%, respectively. There was a statistically significant difference between the two groups according to a log-rank analysis (P = 0.0002). Data are shown in Figure 3.
To allow comparison with older series that used ≤ 0.75 mm as a thickness cut-off point, the SMU data were analyzed to compare survival of patients who had melanomas measuring ≤ 0.75 mm in thickness with the survival of patients who had melanomas measuring 0.76–1.0 mm in thickness. Data are shown in Figure 4. The 10-year survival rate (± SE) for the group with thinner lesions was 96.9% ± 1.3%, compared with 84.3% ± 4.4% for the group with lesions measuring 0.76–1.0 mm in thickness (P ≤ 0.0001).
Patients who had Clark level II or III melanomas were compared with patients who had level IV melanomas. There were no patients with level V melanomas. There was a significant difference in survival between these groups on log-rank analysis, as shown in Figure 5 (P = 0.01). For patients who had level II or III lesions, the survival rate at 10 years was 94.1% ± 1.96% (SE). For patients with level IV lesions, the survival rate at 10 years was 89.0% ± 4.4% (SE).
Deaths from melanoma continued to occur more than 10 years after diagnosis. Both SMU and NSWCCR data showed a continuing decline in melanoma-specific survival after 10 years.
On multivariate analysis, thickness, age, ulceration, and mitotic rate were significant predictors of melanoma-specific survival (Table 4). Gender and Clark level did not reach statistical significance.
|Variable||Hazard ratio||95% CI||P value|
Among 2746 SMU patients who were analyzed, there were only 89 deaths. This group was reviewed in more detail, and data on the group are shown in Table 5. The median melanoma thickness in this group was 0.81 mm. Among the 89 patients who died, the presence or absence of ulceration was recorded in 76 patients and was present in 10 patients. Mitotic rate was recorded in 64 of the patients who died, and the median number of mitoses per mm2 was 1. The presence or absence of regression was recorded in 33 patients who died, and among these patients, regression was present in 20. The median time to recurrence was 49.8 months (range, 7.7–192.0 months). The median time from diagnosis to death was 65.9 months (range, 5–197 months).
|Characteristic||No. of patients||Median (range)|
|Age at diagnosis (yrs)||53.7 (16.1–87.5)|
|Mitotic rate (per mm2)||1|
|Time to recurrence (mos)||49.8 (7.7–192.7)|
|Time to death (mos)||65.9 (4.9–197.0)|
Determining a realistic prognosis for the large number of patients with thin primary cutaneous melanomas is important for many reasons, not least of which is reassurance to a newly diagnosed patient that he or she has an excellent chance of cure. It also is crucial that treating physicians have an accurate idea of the outlook for patients with thin melanomas to avoid the temptation to overtreat patients who have a disease with an excellent prognosis or to undertreat a subset of patients who have a poorer prognosis. To our knowledge, the current study is the largest yet reported on patients with melanomas measuring ≤ 1 mm in thickness. To obtain as accurate a prognostic estimate as possible for this group of patients, two separate data sets—one population-based and the other clinic-based—were analyzed.
The rationale for using 1.0 mm as a cut-off point to define thin melanomas has been explained elsewhere.1 The effect is that, with some exceptions, all patients in this group are treated in a similar fashion and undergo the same surgical excision, staging, entry into adjuvant therapy trials, and follow-up.14–16 Ulceration is uncommon in thin melanomas, but when it is present, it confers a poorer prognosis. This finding is reflected by the presence of a separate category for ulcerated thin melanomas in the new AJCC staging system.1 Today, patients with these types of lesions often are offered a sentinel lymph node biopsy procedure.
Accurate prognostic estimates for patients with thin lesions are surprisingly difficult to obtain from the literature. Most published figures are from institution-based studies and thus are subject to referral bias.3, 17–21 Referral bias for these types of data sets can occur when patients with thin lesions are referred to a tertiary center only when they develop a problem, i.e., recurrence. Thus, a cohort of patients from a single center, regardless of its size, will be likely to have a poorer survival compared with a group of comparable patients who have lesions of similar thickness from a population-based data set. Many reports provide 5-year survival rates rather than more realistic 10-year survival rates.4, 9, 22 Even 10 years is insufficient to account for all recurrences. Thin melanomas appear to recur relatively late compared with thicker lesions.23–25 Population-based data tend to be less current and may not include what are now understood as important prognostic criteria, such as ulceration.4, 7, 9 These data also may be difficult to interpret, because survival is reported at different cut-off points. The quality of registry data is variable and may tend to overestimate survival because of incomplete or inaccurate follow-up. For example, a death resulting from metastatic melanoma may not be attributed to a thin lesion that was removed many years earlier.
The largest analysis of prognosis according to microstaging was by Balch et al., who used the analysis to formulate the new AJCC staging system.2 The data used by the AJCC Staging Committee came from 17,600 patients from 13 cancer centers and groups. Of the 17,600 patients analyzed, 8592 were from the SMU. The 10-year actuarial survival for patients from the SMU who had T1a lesions was 86.2% (McKinnan JG, Thompson JF, Colman MN, 2002; unpublished data). Using a parametric survival model, Gamel et al. recently published a report of a single-center study that estimated that the chance of cure for patients with melanomas measuring ≤ 1 mm in thickness ranged from 57.1% to 80.0%, depending on the patient's subgroup.21 It seems likely that referral bias may have produced artificially low survival rates in these studies. Published survival data from several centers are summarized in Table 6.
|Study||Location||No. of patients||Thickness group||10 yr survival (%)|
|Balch et al., 20012||Multiple||4510||≤ 1.0 mm, nonulcerated||87.9|
|1380||≤ 1.0 mm, ulcerated||83.1|
|Buttner et al., 19953||Germany||1849||≤ 1.0 mm||94.5|
|Morton et al., 199317||United States||749||≤ 0.75 mm||96.0|
|Averbook et al., 200218||United States||677||≤1.59 mm nonulcerated||97.6|
|50||≤ 1.59 mm ulcerated||79.7|
|Schuchter et al., 199619||United States||206||≤ 0.75 mm||96.0|
MacKie et al. published survival statistics from the Scottish Melanoma Registry, which contained 6288 patients.7 Results were collected for patients with lesions measuring ≤ 1.5 mm thick and were reported as actuarial survival at 10 years. The survival rate was 88% for males and 92%. for females. Stang et al. reported cancer registry data from Saarland in Germany.5 In their data, there were 361 patients with melanomas measuring ≤ 1.5 mm thick. The 5-year survival rate was 91% for males and 95% for females. Barnhill et al. reported results from Connecticut, but their numbers were relatively small, and survival was reported for only 5 years.9 The South Australia Cancer Registry has published population-based data from 1977 to 1999 for a large number of patients with thin melanomas.11 Survival data from population-based reports are summarized in Table 7. Although, theoretically, they are more accurate than reports from tertiary treatment centers, population-based reports are not free of problems either. Survival rates are highly dependent on the accuracy of both follow-up data and reported cause of death. Furthermore, deaths may be under reported because of migration out of the country. Population-based reports usually do not include data on prognostic features other than thickness.
|Study||Location||No. of patients||Thickness group||Melanoma specific survival rate (%)|
|5 yr||10 yr|
|MacKie et al., 19977||Scotland||6288a||Male; < 1.5 mm||—||88.0|
|Female; < 1.5 mm||—||92.0|
|South Australian Cancer Registry, 199911||Australia||2737||≤ 0.5 mm||—||98.0|
|Stang et al., 20015||Germany||361||Male; ≤ 1.5 mm||91.0||—|
|Female; ≤ 1.5 mm||95.0||—|
|Heenan et al., 199110||Australia||238||≤ 0.75 mm||98.0||—|
|Barnhill et al., 19969||United States||240||≤ 0.75 mm||97.0||—|
New South Wales Population Data
The NSWCCR records demographic and mortality data for all patients with a cancer diagnosis in New South Wales. It receives a diagnosis from the pathologist interpreting the biopsy and a follow-up report from the treating physician or notification of death from the Death Certificate file from the State Registrar of Births, Deaths, and Marriages and the National Death Index file. Diagnoses are updated in the registry according to the most recent pathologic review. Furthermore, the NSWCCR receives notification of all deaths in Australia, even if they occur outside the State of New South Wales. For the current study, although the number of patients is large, follow-up may be inadequate for detecting late mortalities, given the long natural history of thin melanoma. The reported survival rate of 97.9%, therefore, may be overly optimistic; however, it is fairly close to the figure reported by the South Australian Registry.
Although the SMU is a single center, there are reasons to believe that survival data for its patients may provide an accurate representation of the natural history of thin melanomas. Both pathologic interpretation and treatment are of high quality and consistency, and, most important, follow-up is careful and thorough. All SMU patients who were included in the current study were followed. Data were collected prospectively, and the center had the largest experience in the world, with information on over 20,000 patients recorded. We attempted to reduce referral bias by including only those patients who received definitive treatment at the SMU in the analysis. However, the mean thickness of the melanomas in patients from the SMU was slightly greater than the mean thickness of melanomas in patients the NSWCCR (0.61 mm vs. 0.55 mm, respectively), suggesting that a referral bias still exists. In contrast, passive follow-up of patients from the NSWCCR group may have failed to identify a small number of patients who died of melanoma, particularly if they left Australia prior to death, resulting in a falsely high survival estimate. Therefore, it seems likely that the true long-term survival of patients with thin melanoma lies somewhere between the two different estimates obtained in this study.
Analysis of prognostic variables was conducted based on the SMU data set. Data from the NSWCCR were not sufficiently detailed to allow a similar analysis. Many studies have demonstrated that ulceration is an important and independent prognostic factor.2, 26–28 Multivariate analysis of our data confirmed that the presence of ulceration confers a significantly poorer prognosis on patients with thin melanomas, with a 10-year Kaplan–Meier survival rate of only 84% (P = 0.004). However, the sample size was relatively small (n = 92), because ulceration of thin melanomas is not a common event. The 96.8% survival rate for the group of patients with lesions measuring ≤ 0.75 mm in thickness appears to be comparable to other series and to population-based data.3, 11, 17–19 Survival for the group of patients who had lesions that measured 0.76–1.0 mm in thickness (84.3% at 10 years) was significantly worse compared with the survival of patients who had melanomas measuring ≤ 0.75 mm in thickness on both univariate analysis (P < 0.0001) and multivariate analysis (P = 0.002). Other groups have reported that the Clark level remains an important prognostic criterion for melanomas measuring ≤ 1 mm in thickness.2, 29 Our data confirm that on univariate analysis, there is a significant difference in survival at 10 years between patients with Clark level II–III lesions and patients with Clark level IV lesions. However, that significance is lost on multivariate analysis (P = 0.09). The lack of significance of the Clark level in our data may reflect a smaller sample size or, possibly, the inclusion of mitotic rate, which was not available in the AJCC data set, in the analysis. Azzola et al. reported the independent predictive value of mitotic rate in a recent publication from the SMU.30 The data from the current study confirm that mitotic rate is an independent predictor of survival in patients with thin melanomas.
Data on the 89 patients who died of disease were examined more closely. Not surprisingly, this subset of patients had other negative prognostic features, as outlined in Table 5. The tumors in this group were thicker on average and included 10 ulcerated tumors. The presence or absence of regression was recorded in only 33 of the 89 patients who died, but of these 33, regression was noted in 20. Of the 89 patients who died, only 15 had melanomas that were not ulcerated, showed no regression, had Clark level less than or equal to III, and measured ≤ 0.76 mm in thickness. The reason for the spread of disease and subsequent death in this group remains unclear.
Is Aggressive Treatment Warranted for Thin Melanomas?
The results of the current study suggest that it is unwarranted to exclude all patients with thin melanomas from aggressive investigation and treatment on the basis of population-based survival figures. The current study population is not homogeneous, and there is a wide range of projected survival data. For example, if only patients who had melanomas that were both Level IV and ulcerated (n = 20) are considered, it is found that 6 patients died at a median follow-up of 42.4 months (mortality rate, 30%). The current study adds support to the current SMU policy of offering sentinel lymph node biopsy to patients with melanomas measuring ≤ 1 mm if they have 1 or more poor prognostic factors.
Prognostic estimates based on the AJCC staging system probably are too pessimistic for patients with melanomas measuring ≤ 1 mm thick. Two separate methods of estimating the 10-year survival rate suggest that it is at least 93%. There are identifiable subsets of patients with thin melanomas who have significantly poorer prognoses. These patients may warrant consideration for aggressive investigation and treatment.
The authors thank Ms. Marjorie Colman and Dr. Helen Shaw for their invaluable assistance and the staff of the New South Wales Central Cancer Registry for providing the data used in the analysis.
- 11South Australian Cancer Registry. Epidemiology of cancer in South Australia. Adelaide: Department of Human Services, 2000: 120–121.
- 22The National Cancer Data Base report on cutaneous and noncutaneous melanoma: a summary of 84,836 cases from the past decade. The American College of Surgeons Commission on Cancer and the American Cancer Society. Cancer. 1998; 83: 1664–1678., , .