Ann Surg Oncol (2012) 19:1782–1789 DOI 10.1245/s10434-012-2265-y
ORIGINAL ARTICLE – MELANOMAS
Clinical and Pathologic Factors Associated with Distant Metastasis and Survival in Patients with Thin Primary Cutaneous Melanoma Rajmohan Murali, MBBS, MD, FRCPA1,2,3, Lauren E. Haydu, BSCHE, MIPH2,4, Georgina V. Long, MBBS, PhD, FRACP2,5, Michael J. Quinn, MBBS, FRACS2,4, Robyn P. M. Saw, MBBS, FRACS2,4, Kerwin Shannon, MBBS, FRACS2,4, Andrew J. Spillane, MD, FRACS2,4, Jonathan R. Stretch, MBBS, DPhil, FRACS2,4, Richard F. Kefford, MBBS, FRACP2,5,6, John F. Thompson, MBBS, MD, FRACS, FACS2,4, and Richard A. Scolyer, MBBS, MD, FRCPA, FRCPath1,2,3 Tissue Pathology and Diagnostic Oncology, Royal Prince Alfred Hospital, Sydney, NSW, Australia; 2Melanoma Institute Australia, Sydney, NSW, Australia; 3Discipline of Pathology, Sydney Medical School, The University of Sydney, Sydney, NSW, Australia; 4Discipline of Surgery, Sydney Medical School, The University of Sydney, Sydney, NSW, Australia; 5 Discipline of Medicine, Sydney Medical School, The University of Sydney, Sydney, NSW, Australia; 6Westmead Institute for Cancer Research, Westmead Hospital, Westmead, NSW, Australia 1
ABSTRACT Background. Approximately 3–5% of patients with thin (B1 mm) cutaneous melanomas develop distant metastases. We sought to identify clinical and pathologic factors associated with distant metastasis and survival in a large number of patients with thin melanoma treated at a single institution. Methods. We identified patients with a single invasive melanoma B1 mm in thickness diagnosed between January 1983 and December 2003 who developed distant metastasis (cases), and matched patients with no recorded recurrence during follow-up (control subjects). Cases and control subjects were matched for age, sex, and year of primary melanoma diagnosis. Associations of clinical and pathologic parameters with distant metastasis-free survival and melanoma-specific survival were analyzed. Results. A total of 178 cases and 178 control subjects were identified. Factors associated with development of distant metastasis were: increasing Breslow thickness
(P \ 0.001), increasing Clark level of invasion (P \ 0.001), increasing mitotic rate (P = 0.001), ulceration (P = 0.025), and American Joint Committee on Cancer T subcategory (P \ 0.001). Multivariable models including Breslow thickness (but not Clark level) showed that factors independently associated with poorer distant metastasis-free survival were increasing age [hazard ratio (HR) 1.01, 95% confidence interval (CI) 1.00–1.02]; increasing Breslow thickness (HR 3.21, 95% CI 1.73–5.94, and HR 3.77, 95% CI 2.11–6.74 for 0.51–0.75 mm and 0.76–1.00 mm, respectively, compared with 0.01– 0.50 mm); ulceration (HR 1.87, 95% CI 1.14–3.06) and mitotic rate (HR 1.13, 95% CI 1.05–1.21). Similar associations with melanoma-specific survival were found. Conclusions. Clinical and pathologic predictors of distant metastasis and survival identified in this large study of patients with thin primary cutaneous melanomas will enable more accurate stratification of risk of distant metastasis and poor survival in such patients, and will assist in formulating clinical management and follow-up regimens based on the level of risk.
Rajmohan Murali and Lauren E. Haydu contributed equally to this study. Ó Society of Surgical Oncology 2012 First Received: 10 October 2011; Published Online: 14 February 2012 R. Murali, MBBS, MD, FRCPA e-mail:
[email protected]
The risk of developing distant metastasis in patients with primary cutaneous melanoma increases with increasing primary tumor thickness. Most newly diagnosed patients with melanoma have thin (B1 mm) tumors.1–3 Distant metastasis occurs in approximately 3–5% of patients with thin melanoma.4,5
Metastasis and Survival in Thin Melanoma
Previous studies investigating clinical and pathologic parameters associated with survival and distant metastasis in patients with thin melanoma have produced conflicting findings with respect to the relative predictive value of individual parameters. Factors such as increasing age, male sex, anatomic location on head/neck/trunk, Breslow thickness [0.75 mm or [0.8 mm, presence of mitoses, vertical growth, Clark level, ulceration, and regression have been shown in some studies to be significantly associated with regional node involvement and metastatic potential.2–10 However, other studies have suggested that these parameters (with the exception of Breslow thickness, vertical growth, and the presence of mitoses) have little or no significant association with metastasis and survival in patients with thin melanomas.2,3,6,9,10 The contradictory findings are probably due a combination of the low frequency of metastasizing thin melanomas, the poor reproducibility of assessment of some pathologic parameters, and differences in study populations and study design. In this case-control study, we sought to identify clinical and pathologic factors associated with the development of distant metastasis and survival in a large number of patients with thin primary cutaneous melanoma who were managed at a single institution in accordance with recommended clinical management guidelines.11 METHODS The study was carried out with the approval of the ethics committee of the Sydney South West Area Health Service. Patients with a single invasive primary cutaneous melanoma B1 mm in thickness and diagnosed between January 1983 and December 2003 were identified from the Melanoma Institute Australia (MIA) database. All patients were followed up at MIA. Within this cohort, patients who developed distant metastasis (defined as recurrence beyond the regional lymph node field) during the follow-up period (cases) were identified, together with patients without recorded recurrences during the follow-up period (control subjects). For the purposes of this study, involvement of second echelon nodes (iliac for lower extremity melanomas, subclavian for upper extremity melanomas) was considered stage III, not stage IV, disease. For each case, one control subject, matched for sex, age at diagnosis (within 2 years), year of diagnosis (within 3 years), and having longer follow-up, was selected. The primary tumors were centrally reviewed by MIAaffiliated pathologists at the Royal Prince Alfred Hospital, Sydney, Australia. Ulceration was defined as full-thickness loss of epidermis overlying the invasive melanoma with associated fibrinous exudates and/or stromal response. Regression was categorized into three stages:
1783
early (tumor-infiltrating lymphocytes), intermediate and late. Complete or late regression was characterized by an area devoid of melanoma in the epidermis and dermis with fibrosis, often flanked on one or both sides by residual melanoma. The epidermis was frequently attenuated, with loss of the rete ridges, and the underlying dermis showed angiofibroplasia. In intermediate stage regression there were often a few associated lymphocytes, variable numbers of melanophages, some edema and telangiectasia. Clinical, pathologic, and follow-up data for the patients were retrieved from the MIA database. The associations of clinical and pathologic parameters with distant metastasisfree survival (DMFS) and melanoma-specific survival (MSS) were analyzed. All analyses were carried out by IBM SPSS Statistics 19.0 software (IBM, Armonk, NY). Wilcoxon signed rank test, McNemar’s test, and conditional multinomial logistic regression were employed to measure statistical associations. DMFS was defined as the interval between diagnosis of primary melanoma and diagnosis of distant metastasis. MSS was defined as the interval between diagnosis of primary melanoma and death from melanoma. Patients with no events (i.e., no distant metastasis) during the follow-up period were censored. Survival analysis was carried out by the Kaplan-Meier method. Differences between survival functions for different strata were assessed with the Mantel–Cox log rank statistic. P values were adjusted by the Bonferroni correction for multiple subgroup comparisons. Multivariable analyses of prognostic factors were carried out by the Cox proportional hazard models. In all multivariate models, only one of either Breslow thickness or Clark level of invasion was included in order to avoid collinearity issues. P values of \0.05 were considered to be statistically significant.
RESULTS 5,628 patients with single thin primary melanomas were identified, of whom 178 (3.2%) developed distant metastases. The cohort for analysis in this study included the 178 patients who developed distant metastasis (cases) and 178 patients who did not develop a recurrence (control subjects) during the follow-up period. The case and control subjects were matched for sex; for age in years at diagnosis [exact match in 93 (52.2%), ±1 year in 70 (39.3%), and ±2 years in 15 (8.4%) patients]; and for year of diagnosis [exact match in 135 (75.8%), ±1 year in 19 (10.7%), and ±2 years in 20 (11.2%), and ±3 years in 4 (2.2%) patients]. Follow-up from the date of diagnosis of primary melanoma for each control was always greater than that of the corresponding matched case, and the difference
1784
R. Murali et al.
(n = 62, 34.8%), distant lymph nodes, skin or soft tissue (n = 78, 43.8%), visceral organs other than brain (n = 81, 45.5%), and brain (n = 36, 20.2%). At first presentation of distant metastasis, 46 patients (25.8%) were subcategory M1a, 39 (21.9%) were subcategory M1b, and 93 (52.2%) were subcategory M1c [American Joint Committee on Cancer (AJCC) melanoma staging classification, 7th edition, 2009].12 Factors associated with development of distant metastasis during the follow-up period are presented in Table 1. Statistically significant factors included the presence of ulceration, increasing number of dermal mitoses, Clark level of invasion and increasing Breslow thickness, and AJCC 7th edition T subcategory (T1b vs. T1a) (Fig. 1). In a conditional logistic regression model testing the differences of Breslow thickness, Clark level of invasion, dermal
ranged from 1 to 282 months with a median difference of 55 months. The mean age at diagnosis of primary melanoma was 49 years (range 17–81 years), and 122 patients (34.3%) were female. Mean and median follow-up duration were 123.4 and 111.0 months, respectively. During the followup period, 140 patients died of melanoma, all of whom belonged to the group that developed distant metastases. Median overall survival for patients who developed distant metastases was 79.0 months [95% confidence interval (CI) 62.8–95.2], and was not reached for the group that did not develop distant metastases. First distant metastases were present in 1 anatomic site in 134 patients (75.3%), 2 sites in 28 patients (15.7%), 3 sites in nine patients (5.1%), or [3 sites in seven patients (3.9%). The sites of first distant metastasis included lung
TABLE 1 Associations of clinical and pathologic factors with development of distant metastasis Characteristica
Distant metastasis
No distant metastasis
%b
n
P value and conditional odds ratioc
Matched
Age at primary diagnosis, y B50
87
88
50%
173
[50
91
90
50%
183
117
117
50%
234
61
61
50%
122
Patient sex Male Female
Matched
Site of primary tumor Head and neck
27
24
53%
51
Extremity
56
66
46%
122
Trunk
95
88
52%
183
89
100
47%
189
14
2
88%
16
Histopathologic subtype Superficial spreading Nodular Desmoplastic
2
5
29%
7
Lentigo maligna melanoma
6
9
40%
15
Mixed/other
6
2
75%
8
Missing data
61
60
50%
121
P = 0.679
P = 0.787
Breslow thickness Median, mm
0.78
0.55
–
356
P \ 0.001
0–0.5 mm
34
79
30%
113
P \ 0.001
0.51–0.75 mm
54
56
49%
110
0.76–1.00 mm
90
43
68%
133
Clark level of invasion II
32
83
28%
115
III
91
57
61%
148
IV Missing data
43 12
33 5
57% 70%
76 17
P \ 0.001
Dermal mitoses Mitotic rate per mm2 (median) Mitoses absent
1
0
50
63
44%
262
P = 0.001
113
P = 0.106
Metastasis and Survival in Thin Melanoma
1785
TABLE 1 continued Characteristica
Distant metastasis
No distant metastasis
%b
n
P value and conditional odds ratioc
Mitoses present
91
58
61%
149
Missing data
37
57
39%
94
Ulceration Absent
134
141
49%
275
Present
23
6
79%
29
Missing data
21
31
40%
52
P = 0.025, OR 3.0, 95% CI 1.14–9.23
Regression Absent
41
31
57%
72
Early
45
60
43%
105 17
Intermediate
7
10
41%
Late
20
13
61%
33
Missing data
65
64
50%
129
P = 0.787
AJCC 7th edition T subcategory (2009) P \ 0.001, OR 2.4, 95% CI 1.4–4.1
T1a
56
83
40%
T1b
102
65
61%
20
30
40%
169 10
171 8
50% 56%
340 18
T1 (not tested) Sentinel lymph node biopsy Not performed Performed
P = 0.617
Sentinel lymph node status Negative
10
7
59%
17
Positive
0
1
0%
1
Not computed
OR odds ratio a
‘‘Missing data’’ indicates that data were not recorded in the Melanoma Institute Australia database
b
Percentage of patients who developed distant metastasis
c
McNemar’s test for dichotomous variables, Wilcoxon signed rank test for all others
mitoses, and ulceration for matched pairs (n = 87), increasing Breslow thickness remained the only significant variable influencing the risk of distant metastasis (odds ratio 22.1, 95% CI 2.5–193.1, P = 0.005). There was a nonsignificant trend in association of the presence of dermal mitoses with risk of distant metastasis (odds ratio 1.2, 95% CI 1.0–1.4, P = 0.060); however, data on dermal mitotic rate were unavailable (in one or both tumors) for 79 of 178 matched pairs. In univariate analysis of the overall cohort, time to development of distant metastasis was significantly influenced by the presence of ulceration, presence of dermal mitoses, Breslow thickness, Clark level of invasion, and AJCC T subcategory (7th edition, 2009) at diagnosis of primary melanoma (Fig. 1). In 2 multivariate models including either Breslow thickness or Clark level, respectively, factors independently associated with both poorer DMFS and with poorer MSS were increasing tumor depth (Breslow thickness or Clark level in each model,
respectively), presence of ulceration, increasing dermal mitoses and increasing age (Table 2). In the subgroup of patients that developed a distant metastasis (n = 178), the only factor found to significantly influence overall survival after the development of first metastasis was the M subcategory at first presentation of distant metastasis. Patients diagnosed with metastatic melanoma in their visceral organs had poorer survival compared with those with distant skin, soft tissue or lymph node metastases (P = 0.015) (Fig. 2). There was no significant difference when comparing lung with other visceral organs or lung with skin, soft tissue or lymph node metastases.
DISCUSSION Thickness of primary tumor is a key determinant of prognosis in cutaneous melanoma patients, and the great majority of patients with thin (B1 mm) melanomas have an
1786
R. Murali et al.
a
b
Cumulative survival 1.0
Cumulative survival 1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0
1
2
3
4
5
6
7
8
9
10
11
0
1
Distant metastasis free survival (years) Breslow thickness 0.00–0.50 mm A vs. B 0.51–0.75 mm A vs. C 0.76–1.00 mm B vs. C
2
3
d
Cumulative survival 1.0
Cumulative survival 1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
1
2
3
4
5
6
7
8
5
6
7
8
9
10
11
10
11
Clark level of invasion Clark II A vs. B p < 0.001 Clark III A vs. C p < 0.001 Clark IV B vs. C N.S.
p = 0.009 p < 0.001 p = 0.033
c
0
4
Distant metastasis free survival (years)
9
10
11
Distant metastasis free survival (years)
0
1
2
3
4
5
6
7
8
9
Distant metastasis free survival (years)
Ulceration Not ulcerated p < 0.001 Ulcerated p < 0.001
Mitoses Absent Present
p = 0.008
e Cumulative survival 1.0
0.8
0.6
0.4
0.2
0
1
2
3
4
5
6
7
8
9
10
11
Distant metastasis free survival (years) AJCC 7th edition Tstage T1a p < 0.001 T1b
FIG. 1 Association of clinical and pathologic factors with DMFS. a Breslow thickness. b Clark level. c Ulceration. d Mitoses. e T subcategory (AJCC staging system, 7th edition, 2009)
Metastasis and Survival in Thin Melanoma
1787
TABLE 2 Multivariate analyses of the associations of clinical and pathologic factors with survival Model with Clark level of invasion
DMFS (n = 243) P
MSS (n = 243)
HR
95.0% CI
P
Lower
Upper
1.00
1.03
HR
95.0% CI Lower
Upper
1.02
1.00
1.03
Age at diagnosisa
.010
Clark II (reference)
\0.001
Clark III
\0.001
3.49
2.04
5.95
.002
0.34
0.17
0.67
Clark IV
\0.001
2.89
1.62
5.15
.309
1.26
0.80
1.99
Ulceration
.005
2.04
1.24
3.34
.003
2.25
1.31
3.87
Mitotic ratea
\0.001
1.13
1.06
1.21
.013
1.11
1.02
1.20
Model with Breslow thickness
1.01
.010 \0.001
DMFS (n = 245) P
MSS (n = 245)
HR
95.0% CI
P
Lower
Upper
1.00
1.02
HR
95.0% CI Lower
Upper
1.02
1.00
1.03
Age at diagnosisa
.022
0–0.5 mm (reference)
\0.001
0.51–0.75 mm
\0.001
3.21
1.73
5.94
.001
3.36
1.64
6.87
0.76–1.00 mm
\0.001
3.77
2.11
6.74
\0.001
3.83
1.93
7.60
Ulceration
.014
1.87
1.14
3.06
.010
2.05
1.19
3.53
Mitotic ratea
.001
1.13
1.05
1.21
.017
1.10
1.02
1.19
a
1.01
.022 .001
Continuous variable
HR hazard ratio derived from Cox proportional hazard models, 95% CI 95% confidence interval for HR Cumulative survival 1.0
0.8
0.6
0.4
0.2
0
1
2
3
4
5
6
Overall survival from distant metastasis (years) Mstage Skin soft tissue lymph node Lung Other visceral organs
A vs. B A vs. C B vs. C
N.S. p = 0.015 N.S.
FIG. 2 Association of M subcategory with overall survival from distant metastasis
excellent outcome. However, a minority of patients with thin melanomas do recur locally, in regional lymph nodes, or at distant anatomic locations.4,5 Metastasis, particularly to distant sites, is an established marker of tumor progression to an advanced stage, and is associated with poor outcomes. Therefore, comparative study of two groups of patients with thin melanomas, namely those in whom
distant metastasis occurred, and those in whom there was no evidence of disease progression, is likely to reveal clinical and pathologic factors that are most predictive of adverse outcomes. The strengths of the present study are the inclusion of a large number of matched patients with metastatic and nonrecurrent thin melanomas, as well as the uniform clinical management and pathologic review, respectively, of patients and their tumors at a single specialist melanoma treatment center. Only one of either Breslow thickness or Clark level of invasion was used as a measure of tumor depth in multivariate models because they are closely related parameters. Increasing Breslow thickness and increasing Clark level were both independently associated with increased risk of occurrence of distant metastasis, and with poorer DMFS and MSS, as shown in previous studies.2,5,6 It is well documented that assessment of Clark level is less reproducible than assessment of Breslow thickness, and as a consequence, the latter is a more reliable prognostic parameter in thicker melanomas.12–14 Our results suggest that Clark level of invasion may be a more useful predictor of prognosis in thin melanomas than it is in intermediate/thick melanomas. Increasing patient age, increasing tumor thickness, presence of ulceration and mitotic rate were independently associated with poorer DMFS and MSS. Tumor depth and mitotic rate have been shown in numerous studies to be
1788
accurate prognostic indicators and independent predictors of survival in patients with melanoma of any thickness.2–6,12,15–17 The fact that these parameters are also strong predictors of metastasis and survival in thin melanomas indicates that they are markers of aggressive behavior in melanoma. The presence of ulceration is an independent adverse prognostic parameter in thick melanomas, but its predictive value has been more controversial in thin melanomas, with some studies showing that the presence of ulceration is associated with metastatic potential, while others have found no such association.2–6,10,12 This discrepancy may be related to the relative rarity of ulceration in thin melanomas; it has previously been shown that the incidence of ulceration in melanomas is proportional to tumor thickness.15,16 Indeed, ulceration was identified only in 9.5% (29 of 304) of the patients in the present study. Furthermore, it can be very difficult for pathologists to distinguish tumor ulceration from traumatic ulceration (the latter is presumed to have little prognostic relevance, in contrast to the former). Previous studies have reported either a significant association or a lack of association between increasing age and the development of distant metastasis and poor survival in patients with thin melanomas.2,3,6,9,15 In our study, increasing patient age was an independent predictor of DMFS and MSS. Anatomic location of the primary melanoma in the head/ neck region and on the trunk (compared to the limbs) was associated with a slightly increased risk of distant metastasis in patients with thin melanomas (Table 2), but this association was not statistically significant. An association between head/neck location of thin melanoma and poorer survival has been shown in some previous studies, but not in others.3,6,9,10 The papillary dermis is quite thin in the head/neck region and some truncal sites, so that even thin melanomas at these locations extend to the junction of the papillary and reticular dermis or into the reticular dermis. It is possible that the consequent ready access of the tumor to the dermal vascular plexus might lead to increased risk of systemic dissemination. Regression in thin melanomas has been shown in some studies to be significantly associated with regional nodal involvement and metastatic potential, but not in others, including the present study.3,6–9 It is possible that thin melanomas with regression were actually thicker tumors in which the measured thickness of residual melanoma in the regressed tumor is an underestimate of the original thickness of the tumor. In such cases, the metastatic potential of the tumor might be better predicted by the original thickness rather than the thickness of residual melanoma in the regressed tumor, and this might explain the association of metastasis with regression seen in some studies. Furthermore, it is well documented that the histologic assessment
R. Murali et al.
of regression is poorly reproducible, which is likely to be a result, at least in part, of the lack of clear and generally applied criteria for its recognition.18,19 These factors very likely explain the discrepancies in the reported prognostic value of regression in thin melanomas. In summary, we have identified independent clinical and pathologic predictors of distant metastasis and survival in a large cohort of patients with thin primary cutaneous melanomas. These factors may be used to more accurately stratify the risk of distant metastasis and poor survival in patients with thin melanomas, and to optimize clinical management and follow-up regimens by improved estimation of the level of risk. ACKNOWLEDGMENT G.V.L. and R.A.S. are supported by the Cancer Institute New South Wales Clinical Fellowship program. We acknowledge the support of the Cancer Institute New South Wales, the Australian National Health and Medical Research Council, and colleagues at Melanoma Institute Australia and the Department of Tissue Pathology and Diagnostic Oncology, Royal Prince Alfred Hospital.
REFERENCES 1. Reintgen C, Shivers S, Reintgen M, et al. The changing face of malignant melanoma. J Surg Oncol. 2010;101:443–6. 2. McKinnon JG, Yu XQ, McCarthy WH, et al. Prognosis for patients with thin cutaneous melanoma: long-term survival data from New South Wales Central Cancer Registry and the Sydney Melanoma Unit. Cancer. 2003;98:1223–31. 3. Gimotty PA, Guerry D, Ming ME, et al. Thin primary cutaneous malignant melanoma: a prognostic tree for 10-year metastasis is more accurate than American Joint Committee on Cancer staging. J Clin Oncol. 2004;22:3668–76. 4. Kalady MF, White RR, Johnson JL, et al. Thin melanomas: predictive lethal characteristics from a 30-year clinical experience. Ann Surg. 2003;238:528–35. 5. Mansson-Brahme E, Carstensen J, Erhardt K, et al. Prognostic factors in thin cutaneous malignant melanoma. Cancer. 1994;73: 2324–32. 6. Leiter U, Buettner PG, Eigentler TK, et al. Prognostic factors of thin cutaneous melanoma: an analysis of the central malignant melanoma registry of the German Dermatological Society. J Clin Oncol. 2004;22:3660–7. 7. Slingluff CL Jr, Seigler HF. ‘‘Thin’’ malignant melanoma: risk factors and clinical management. Ann Plast Surg. 1992;28:89–94. 8. Slingluff CL, Jr., Vollmer RT, Reintgen DS, et al. Lethal ‘‘thin’’ malignant melanoma. Identifying patients at risk. Ann Surg. 1988;208:150–61. 9. Guitart J, Lowe L, Piepkorn M, et al. Histological characteristics of metastasizing thin melanomas: a case-control study of 43 cases. Arch Dermatol. 2002;138:603–8. 10. Finley JW, Gibbs JF, Rodriguez LM, et al. Pathologic and clinical features influencing outcome of thin cutaneous melanoma: correlation with newly proposed staging system. Am Surg. 2000; 66:527–31. 11. Australian Cancer Network Melanoma Guidelines Revision Working Party. Clinical Practice Guidelines for the management of melanoma in Australia and New Zealand. Wellington: The Cancer Council Australia and Australian Cancer Network, Sydney and New Zealand Guidelines Group; 2008.
Metastasis and Survival in Thin Melanoma 12. Balch CM, Gershenwald JE, Soong SJ, et al. Final version of 2009 AJCC melanoma staging and classification. J Clin Oncol. 2009;27:6199–206. 13. Murali R, Hughes MT, Fitzgerald P, et al. Interobserver variation in the histopathologic reporting of key prognostic parameters, particularly Clark level, affects pathologic staging of primary cutaneous melanoma. Ann Surg. 2009;249:641–7. 14. Scolyer RA, Shaw HM, Thompson JF, et al. Interobserver reproducibility of histopathologic prognostic variables in primary cutaneous melanomas. Am J Surg Pathol. 2003;27:1571–6. 15. Balch CM, Soong SJ, Gershenwald JE, et al. Prognostic factors analysis of 17,600 melanoma patients: validation of the American Joint Committee on Cancer melanoma staging system. J Clin Oncol. 2001;19:3622–34.
1789 16. Azzola MF, Shaw HM, Thompson JF, et al. Tumor mitotic rate is a more powerful prognostic indicator than ulceration in patients with primary cutaneous melanoma: an analysis of 3661 patients from a single center. Cancer. 2003;97:1488–98. 17. Francken AB, Shaw HM, Thompson JF, et al. The prognostic importance of tumor mitotic rate confirmed in 1317 patients with primary cutaneous melanoma and long follow-up. Ann Surg Oncol. 2004;11:426–33. 18. Corona R, Mele A, Amini M, et al. Interobserver variability on the histopathologic diagnosis of cutaneous melanoma and other pigmented skin lesions. J Clin Oncol. 1996;14:1218–23. 19. Lock-Andersen J, Hou-Jensen K, Hansen JP, et al. Observer variation in histological classification of cutaneous malignant melanoma. Scand J Plast Reconstr Surg Hand Surg. 1995;29:141–8.
