Dr. Gershenwald is a member of the Speakers Bureau for Schering-Plough.
Adjuvant high-dose interferon for cutaneous melanoma is most beneficial for patients with early stage III disease
Article first published online: 4 MAR 2008
Copyright © 2008 American Cancer Society
Volume 112, Issue 9, pages 2030–2037, 1 May 2008
How to Cite
Anaya, D. A., Xing, Y., Feng, L., Huang, X., Camacho, L. H., Ross, M. I., Gershenwald, J. E., Lee, J. E., Mansfield, P. F. and Cormier, J. N. (2008), Adjuvant high-dose interferon for cutaneous melanoma is most beneficial for patients with early stage III disease. Cancer, 112: 2030–2037. doi: 10.1002/cncr.23399
- Issue published online: 21 APR 2008
- Article first published online: 4 MAR 2008
- Manuscript Accepted: 12 NOV 2007
- Manuscript Revised: 6 NOV 2007
- Manuscript Received: 17 SEP 2007
- National Cancer Institute Clinical Oncology Research Development Program Award. Grant Number: 5-K12-CA088084
- American Society of Clinical Oncology Career Development Award. Grant Number: 88202
- high-dose interferon;
- adjuvant therapy;
- stage III;
Evidence from randomized trials in the pre-sentinel lymph node biopsy era indicate that adjuvant treatment with high-dose interferon-α (IFN) increases recurrence-free survival (RFS) in patients with high-risk melanoma. However, to the authors' knowledge, the role of this treatment in selected patients with early stage III disease has not been well studied.
The clinical and pathologic characteristics of 486 patients undergoing surgical treatment for stage III melanoma were evaluated and the authors compared outcomes for those given adjuvant treatment with IFN with those patients who had surgery alone. A particular focus was on the effect of IFN therapy on RFS and overall survival (OS) among those patients with stage IIIA disease.
The median follow-up for the entire cohort was 5.2 years; the 5-year RFS and OS rates for the entire group were 41% and 53%, respectively. Adjuvant IFN was given to 141 patients (29%). On multivariate analysis, IFN was found to be the only independent predictor for RFS in patients with stage IIIA disease (hazards ratio of 0.4; 95% confidence interval, 0.2–0.9 [P = .02]). IFN was not found to be associated with increased RFS in patients with more advanced lymph node disease (stage IIIB and stage IIIC). IFN appeared to have no effect on OS in any patient with stage III disease.
Adjuvant treatment with IFN improves RFS in melanoma patients with early stage III disease. The results of the current study should help guide management when considering adjuvant treatment for these patients. Cancer 2008. © 2008 American Cancer Society.
At least 10% of the estimated 59,940 new cases of cutaneous malignant melanoma diagnosed in 2007 will have presented as regional/lymph node (stage III) disease.1–3 The outcome for patients with stage III melanoma varies significantly, with 5-year overall survival (OS) rates ranging from 26% to 67% depending on the extent of metastatic disease (the number of positive lymph nodes and whether the metastasis is microscopic vs macroscopic)2, 3 and primary tumor characteristics (such as the presence of ulceration). In general, surgical treatment has proven inadequate for patients with stage III melanoma, and several adjuvant regimens have been investigated. Although no apparent benefit with regard to OS has been demonstrated, high-dose interferon-α (IFN) has been shown to increase recurrence-free survival (RFS) in patients with high-risk disease,4–11 and to the best of our knowledge, this is the only adjuvant treatment approved by the U.S. Food and Drug Administration (FDA) for malignant melanoma. Nevertheless, IFN is not always recommended because of its significant toxicity profile,10 its effects on quality of life,12, 13 and its cost.14
Recommendations for the use of IFN are derived primarily from the findings of 2 randomized trials (Eastern Cooperative Oncology Group [ECOG] 1684 and Intergroup E1690)4, 5 in which this regimen was compared directly with observation only for patients with thick primary tumors (ie, Breslow thickness of >4 mm) in the absence of lymph node disease (stage IIB) and for patients with primary or recurrent metastatic lymph node disease (stage III).15 The participants in these studies were accrued between 1984 and 1995, primarily before sentinel lymph node biopsy (SLNB) had been introduced; therefore, microscopic lymph node disease was identified in few patients after elective lymph node dissection, which was performed either as a study prerequisite (ECOG 1684) or after selective lymphadenectomy for patients with high-risk disease (Intergroup E1690).
Since that time, SLNB has become the recommended method for the pathologic staging of all primary melanomas measuring >1-mm thick and in selected patients with thin (<1 mm) melanomas.16–18 As a result of the adoption of SLNB19 and the use of enhanced tissue processing (eg, immunohistochemical staining),16, 20–22 patients with early stage III disease, specifically those with micrometastatic lymph node involvement, are being identified more frequently. This shift is now reflected in the most recent (2002) version of the American Joint Committee on Cancer (AJCC) staging system (sixth edition),2, 3, 23 in which lymph node stage is defined not only by the number of positive lymph nodes but also by the extent of lymph node involvement (microscopic vs macroscopic), as derived from tissue evaluation that includes serial sectioning and immunohistochemical analysis of SLNs. Although both of the randomized trials noted above4, 5 included patients with microscopic lymph node disease identified by elective lymph node dissection, few such patients were studied (34 in the ECOG 1684 study and 68 in the Intergroup E1690 study) and the surgical specimens were not subjected to serial sectioning or immunohistochemical staining, which raises the question of whether the outcomes associated with adjuvant IFN can reasonably be extrapolated to patients with micrometastatic lymph node disease as defined by more contemporary methods.
Given the lack of information regarding the role of adjuvant high-dose IFN therapy for patients with early stage III disease, we conducted a retrospective review to identify the impact of IFN therapy on survival outcomes for patients with metastatic lymph node cutaneous melanoma, with the primary objective of examining the subgroup of patients with stage IIIA disease (micrometastatic disease with nonulcerated primary tumor).
MATERIALS AND METHODS
The study was approved by the Institutional Review Board of the University of Texas M. D. Anderson Cancer Center (MDACC). Initially, we identified a total of 804 consecutive patients with melanoma metastatic to ≥1 regional lymph node basins (ie, stage III disease) who had been treated at MDACC between 1990 and 2001. From this group, we selected only those patients treated after 1995 (the time at which high-dose IFN was approved by the FDA as adjuvant therapy for melanoma), yielding the 486 patients who constituted the cohort for this study. Medical records from all patients were reviewed and demographic, clinical, pathologic, treatment, and outcome variables were extracted and recorded. Disease was restaged at the time of review according to the sixth edition of the AJCC staging system.2, 3 All 486 patients had undergone therapeutic lymph node dissection (LND) of the regional lymph node basin or basins at MDACC. Two groups were subsequently identified: those given IFN as adjuvant therapy (IFN group) and those who underwent surgery without adjuvant treatment (observation group) based on recommendations from the treating physician and individual patient preference according to standard protocols described elsewhere.4–6 Patients were followed every 3 months for the first 2 years after LND, at 6-month intervals until Year 5, and yearly thereafter.
RFS time, defined as the period from therapeutic LND to disease recurrence (local/regional or distant) or death, was the primary outcome of interest; OS time, calculated from the date of therapeutic LND to the date of death, was considered as a secondary endpoint. Distant disease-free survival (DDFS) was also calculated as the time from therapeutic LND to the time of distant disease recurrence or death. Patients without disease recurrence or death were censored by using the date of last follow-up for each outcome. The median RFS and OS, as well as the 5-year RFS and OS, were calculated for both the IFN and the observation groups. The association between IFN and RFS was evaluated further for subgroups with pathologic stage IIIA, IIIB, and IIIC disease. Pathologic stage IIIA disease includes patients with up to 3 microscopic lymph node metastases arising from a nonulcerated primary melanoma. Stage IIIB disease consists of patients with up to 3 macroscopic lymph node metastases arising from a nonulcerated primary melanoma, or up to 3 microscopic lymph node metastases arising from an ulcerated primary melanoma, or intralymphatic metastases without lymph node metastases. Stage IIIC disease consists of patients with clinically apparent lymph node metastasis (macroscopic) arising from an ulcerated primary melanoma, or those with ≥4 lymph node metastases, matted lymph node metastasis, or combined lymph node and in-transit metastasis in the absence of distant disease.24
Demographic and clinicopathologic characteristics were compared for patients in the IFN and observation groups using the chi-square test or Fisher exact test, as appropriate. The Kaplan-Meier method was used to estimate RFS and OS in all patients with stage III disease as well as for subsets of patients with stage IIIA, IIIB, or IIIC disease. The Cox proportional hazards model was used to assess the prognostic significance of demographic, clinicopathologic, and treatment variables. A P value of .05 was considered to be statistically significant. All computations were performed using SAS statistical software (version 9.1; SAS Institute, Cary, NC).
The median follow-up time for the 486 patients with stage III melanoma was 5.2 years. Of these patients, 141 (29%) received adjuvant therapy with IFN. The distribution of demographic and clinicopathologic variables for both the IFN and observation groups are listed in Table 1. Patients in the IFN group were more likely to be younger than those in the observation group (54% vs 43% patients age <50 years; P = .02) and less likely to have extracapsular lymph node extension in the LND specimens (12% vs 21%; P = .02). Disease stage was classified according to the sixth edition of the AJCC system in 462 patients; 110 patients (24%) had stage IIIA disease, 192 patients (42%) had stage IIIB disease, and 160 patients (35%) had stage IIIC disease. Disease stage in the remaining 24 patients could not be determined, primarily because of unknown primary tumor characteristics.
|Clinicopathologic factors||No. of patients (%)|
|IFN (n = 141)||Observation (n = 345)||P|
|<50||76 (54%)||147 (43%)|
|≥50||65 (46%)||198 (57%)|
|Female||67 (48%)||141 (41%)|
|Male||74 (52%)||204 (59%)|
|Primary tumor site||NS|
|Head and neck||3 (2%)||8 (2%)|
|Trunk||62 (44%)||132 (38%)|
|Extremities||65 (46%)||183 (53%)|
|Unknown||11 (8%)||22 (7%)|
|Primary tumor histology||NS|
|Superficial spreading||33 (24%)||89 (26%)|
|Nodular||40 (28%)||82 (24%)|
|Acral lentiginous||9 (6%)||24 (7%)|
|Other||48 (34%)||128 (37%)|
|Unknown||11 (8%)||22 (6%)|
|Primary tumor classification||NS|
|T1||14 (10%)||46 (13%)|
|T2||30 (21%)||83 (24%)|
|T3||43 (30%)||99 (29%)|
|T4||36 (26%)||62 (18%)|
|Unknown||18 (13%)||55 (16%)|
|I–III||37 (26%)||84 (24%)|
|IV–V||81 (57%)||194 (56%)|
|Unknown||23 (17%)||67 (20%)|
|Primary tumor ulceration||NS|
|Yes||38 (27%)||99 (29%)|
|No||92 (65%)||219 (63%)|
|Unknown||11 (8%)||27 (8%)|
|Lymph node basin involved||NS|
|Axilla||74 (52%)||172 (50%)|
|Groin||61 (43%)||154 (45%)|
|Other||6 (5%)||19 (5%)|
|Lymph node classification||NS|
|N1||74 (52%)||168 (49%)|
|N2||38 (27%)||87 (25%)|
|N3||29 (21%)||90 (26%)|
|Extracapsular lymph node extension||.02|
|Yes||17 (12%)||71 (21%)|
|No/Unknown||124 (88%)||274 (79%)|
Distant Disease-free Survival
The median DDFS time for the entire cohort (n = 486 patients) was 4.07 years (95% confidence interval [95% CI], 2.75–5.2 years), and the Kaplan-Meier estimated 5-year DDFS for the entire group was 47% (95% CI, 42–52%). The median DDFS time for the IFN group was 5.06 years (95% CI, 3.75 years to not reached) and that for the observation group was 3.02 years (95% CI, 2.22–5.2 years). The Kaplan-Meier estimated 5-year DDFS was 51% (95% CI, 44–61%) for the IFN group and 45% (95% CI, 40–51%) for the observation group (P = .042).
The median RFS time for the entire cohort (n = 486 patients) was 2.27 years (95% CI, 1.84–3.76 years), and the Kaplan-Meier estimated 5-year RFS for the entire group was 41% (95% CI, 36–46%). The median RFS time for the IFN group was 3.82 years (95% CI, 2.21 years to not reached) and that for the observation group was 2.06 years (95% CI, 1.54–2.96 years). The Kaplan-Meier estimated 5-year RFS was 44% (95% CI, 36–54%) for the IFN group and 39% (95% CI, 34–45%) for the observation group (P = .08). Independent prognostic factors for RFS on multivariate analyses were age ≥50 years (hazards ratio [HR] of 1.4; 95% CI, 1.1–1.8 [P = .01]), N2 disease (HR of 1.5; 95% CI, 1.1–2.1 [P = .01]), N3 disease (HR of 3; 95% CI, 2.2–4.1 [P <.001]), and acral lentigenous tumor histology (HR of 1.8; 95% CI, 1.1–2.9 [P = .01]). Receiving adjuvant IFN was not found to be an independent prognostic factor for RFS for the combined group of patients with stage III disease (HR of 0.8; 95% CI, 0.6–1.1 [P = .09]).
The median OS time for the entire stage III cohort was 5.6 years (95% CI, 4.9 years to not reached), and the Kaplan-Meier estimated 5-year OS was 53% (95% CI, 49–58%). The median 5-year OS for the IFN group was 60% (95% CI, 52–69%) and was 50% (95% CI, 45–56%) in the observation group (P = .04). However, when controlling for additional prognostic factors, adjuvant IFN therapy was not found to be an independent prognostic factor of OS for the entire stage III cohort (data not shown).
Effect of Adjuvant IFN According to Disease Substage
Of the 110 patients with stage IIIA disease, 42 (38%) had been given IFN as adjuvant therapy. Of these 42 patients, 12 (29%) developed disease recurrence compared with 27 patients (39%) in the observation group (P = .09 on univariate analysis). The numbers of recurrences between the IFN and the observation groups were more evenly distributed among patients with more advanced disease stage (60% in the IFN group vs 61% in the observation group for stage IIIB disease and 79% in the IFN group vs 74% in the observation group for stage IIIC disease). Unadjusted Kaplan-Meier estimates for DDFS, RFS, and OS were initially examined to evaluate the effect of IFN therapy according to disease substage (Table 2). The 5-year RFS was found to be slightly lower, although not statistically significant, in the observation group (61%) compared with the IFN group (70%) for those with stage IIIA disease (P = .09) (Fig. 1), whereas the 5-year RFS times were found to be similar among patients with stage IIIB disease (36% in the observation group vs 39% in the IFN group) (Fig. 2) or stage IIIC disease (30% in the observation group vs 20% in the IFN group (Fig. 3).
|5-Year survival||95% CI||5-Year survival||95% CI|
When clinicopathologic covariates were accounted for on multivariate analyses, adjuvant IFN therapy was identified as the only independent predictor of RFS for patients with stage IIIA disease, with an HR of 0.4 (95% CI, 0.2–0.9; P = .02). Adjuvant IFN therapy was not found to be an independent predictor for RFS for patients with stage IIIB or IIIC disease (Table 3). Similarly, after controlling for prognostic factors, IFN was not found to be an independent predictor of OS in any of the 3 disease substages (stage IIIA, IIIB, or IIIC) (data not shown).
|Predictor*||Hazards ratio||95% CI||P|
|IFN vs observation||0.4||0.2–0.9||.02|
|Acral lentiginous vs superficial spreading||4.1||1.7–9.8||.002|
|T2 vs T1||2.7||1.2–6||.01|
|T4 vs T1||4.5||1.9–10.9||.001|
|IFN vs observation||0.8||0.5–1.3||.4|
|N3 vs N1||4.3||2–9.3||<.001|
|IFN vs observation||1||0.6–1.6||.9|
|All stage III|
|Age >50 y vs <50 y||1.4||1.1–1.8||.01|
|Acral lentiginous vs superficial spreading||1.8||1.1–2.9||.01|
|N2 vs N1||1.5||1.1–2.1||.01|
|N3 vs N1||3||2.2–4.1||<.001|
|IFN vs observation||0.8||0.6–1.1||.1|
In our covariate-adjusted analysis of 110 melanoma patients with stage IIIA disease, patients who had been treated with adjuvant IFN demonstrated a significant improvement in RFS compared with patients who had not received adjuvant IFN. At 5 years' of follow-up, the use of IFN was associated with lower rates of disease recurrence, with a cumulative absolute benefit of 9%. However, these results did not translate into significant differences in OS. Moreover, this benefit was not evident for patients with more advanced lymph node disease (stages IIIB and IIIC). Patients with stage IIIB disease experienced a similar 5-year RFS regardless of whether they had received IFN. Similarly, no benefit was found for patients treated with IFN who had bulky lymph node disease (stage IIIC), which likely can be explained by the presence of a higher burden of distant micrometastatic disease, which may not derive a clinical benefit from IFN.25
Our study population was a relatively homogeneous group of patients with high-risk, lymph node-positive melanoma, patients who traditionally are considered likely to benefit from adjuvant IFN. The IFN and observation subgroups were relatively balanced with respect to known prognostic factors, except that patients with pathologic evidence of extracapsular lymph node extension and older patients were less likely to have received IFN (P = .02). Because our multivariate model adjusted for these factors in addition to all other known prognostic factors, the effects of IFN were unlikely to have been confounded by these 2 variables. In addition, the current study population was homogeneous relative to the techniques used for pathologic lymph node staging (SLNB) including immunohistochemical analysis. The most recent revision of the AJCC staging system for melanoma (2002) identified both ulceration and micrometastatic versus macrometastatic lymph node disease as important prognostic factors,2, 3, 23 factors that were not part of the staging classification at the time of the previous IFN trials (the ECOG 1684 and Intergroup E1690 studies).4, 5 In the current analysis, patients with micrometastatic lymph node disease and no primary tumor ulceration (stage IIIA) were identified and examined as a separate cohort, thereby allowing us to evaluate the effect of IFN therapy in a specific subgroup that to our knowledge had not been studied previously.
Several points merit comment when comparing our findings with those of published randomized trials of high-dose IFN therapy,4–7 a few systematic reviews, and a meta-analysis on the topic.9–11 First, the randomized trials included heterogeneous populations of patients with high-risk disease with respect to prognosis. Second, the previously published trials included only relatively small numbers of patients with microscopic-only disease, and the relative accuracy with which this condition was diagnosed was lower compared with the current report. For example, in the ECOG 1684 trial, microscopic-only disease was identified after mandatory LND,4 which is recognized to be less accurate for pathologic staging than current SLNB techniques.16, 20, 21 In the Intergroup E1690 trial, LND was not required for patients without clinically apparent lymph node metastasis and was performed electively in only a minority of these patients (24%; selection criteria not specified), allowing for an even greater gap in the accuracy of the diagnosis of microscopic-only disease based on current standards.5 The Intergroup E1694 trial, which compared high-dose IFN therapy versus the ganglioside GM2-keyhole limpet hemocyanin (GMK) vaccine in patients with high-risk melanoma, used the same guidelines for selective lymphadenectomy as those used in the Intergroup E1690 trial.5, 6 To our knowledge, these 3 trials together constitute the largest numbers of patients with high-risk disease (280 patients, 608 patients, and 774 patients, respectively) randomized to high-dose IFN versus observation (the ECOG 1684 and Intergroup E1690 studies) or to high-dose IFN versus other treatment (the Intergroup E1694 study). However, the numbers of patients identified with microscopic lymph node disease in each trial and in a pooled analysis of the ECOG 1684 and Intergroup E1690 studies7 were significantly less than the numbers in our report (Table 4). Furthermore, the presence of micrometastatic disease in relation to ulcerated and nonulcerated primary tumors is not well delineated, limiting the interpretation of their results in different prognostic groups stratified using current standards (specifically, stage IIIA disease).
|Study||AJCC disease stage||Total No. of patients||No. of patients with micrometastatic lymph node disease||Impact of IFN on 5-year RFS|
|Total||IFN||Observation only||Effect on total study population||Primary finding|
|Kirkwood 19964*||IIB, III||280||34||20||14||Absolute increase in RFS of 11% (P = .002); P = .001 after adjustment for multiple variables||Major impact on patients with clinically evident lymph node-positive disease|
|Kirkwood 20005||IIB, III||608||68†||18||29||Absolute increase in RFS of 9% (HR of 1.28; P = .05); P = .03 after adjustment for multiple variables||Major impact on patients with lymph node-positive disease, particularly those with 2–3 positive lymph nodes|
|Kirkwood 19947||IIB, III||713||NR||NR||NR||Increased RFS (HR of 1.3; P = .006); P = .01 after adjustment for multiple variables||No data on subset analysis|
|Kirkwood 20016‡||IIB, III||774||NA§||NA§||NA||Absolute increase in 2-year RFS of 13% (HR of 1.49; P = .0004); P = .0007 after adjustment for multiple variables||High-dose IFN was of the most benefit for patients with no lymph node involvement (P = .01).|
|Current study||III||486||110||42||68||Absolute increase in RFS of 5% (P = .08); P = .1 after adjustment for multiple variables||Stage IIIA absolute increase in RFS of 9% (P = .09); P = .02 after adjustment for multiple variables|
The effects of adjuvant high-dose IFN therapy in patients with microscopic-only lymph node disease, and stage IIIA disease specifically, are inconsistent in the published literature. For example, in the ECOG 1684 study, the benefit with regard to RFS was noted in all patients with lymph node-positive disease, although the effect was more pronounced in those with clinically positive lymph nodes.4 In the Intergroup E1690 study, patients with 2 or 3 positive lymph nodes were noted to derive the most benefit with respect to RFS after treatment with high-dose IFN.5 In contrast, in the Intergroup E1694 study, patients with high-risk, lymph node-negative disease (stage IIB) benefited the most from adjuvant IFN therapy.6 This group of high-risk patients, which likely included a significant proportion of patients with unidentified micrometastatic disease, may well represent the group that is most similar to our contemporary cohort. These differences in survival outcomes previously have been explained by the heterogeneity of each study population and by the differences in patient characteristics between trials. Furthermore, the different distribution and small numbers of patients with microscopic-only disease in these trials may also contribute to these seemingly contradictory results.
Although the patient population in the current analysis was homogeneous and the analyses were adjusted for known prognostic factors, the interpretation of these findings must take into consideration that the study size was relatively small and that patients were not randomized between the IFN and the observation groups. Although the Cox model can adjust for known clinicopathologic and treatment factors, we cannot exclude the possibility that the effects of IFN arose as a consequence of an imbalance in unknown confounding factors. In addition, the retrospective nature of this analysis did not allow us to evaluate the toxicity of IFN and whether dose adjustments may have influenced the final results.
In summary, this retrospective analysis of patients with lymph node-positive melanoma suggests that the use of adjuvant IFN may confer a benefit in terms of RFS and that such a benefit is most striking for those patients with early, –micrometastatic- lymph node disease. Despite the ongoing debate regarding the risk-benefit ratio of adjuvant IFN, the current study data support the belief that patients with stage IIIA melanoma should be informed that IFN may have better clinical outcomes than reported in previous trials, which had included patients with a greater metastatic lymph node tumor burden and more advanced stage III disease.
We thank Christine Wogan for editorial assistance and Debbie Dunaway for article preparation
- 6High-dose interferon alfa-2b significantly prolongs relapse-free and overall survival compared with the GM2-KLH/QS-21 vaccine in patients with resected stage IIB-III melanoma: results of intergroup trial E1694/S9512/C509801. J Clin Oncol. 2001; 19: 2370–2380., , , et al.
- 20Interim results of the multicenter selective lymphadenectomy trial (MSLR-I) in clinical stage I melanoma. Presented at the American Society of Clinical Oncology Annual Meeting, Orlando, FL, May 13–17, 2005., , , , .
- 21Lymphatic mapping and sentinel lymphadenectomy for early-stage melanoma: therapeutic utility and implications of nodal microanatomy and molecular staging for improving the accuracy of detection of nodal micrometastases. Ann Surg 2003; 238: 538–549; discussion 549–550., , , et al.
- 24Melanoma of the skin. In: GreeneFL,PageDL,FlemingID, et al., eds. AJCC cancer staging manual.6th ed. New York: Springer; 2002: 239.