Long-term Results of Hyperthermic, Isolated Limb Perfusion for Melanoma

Abstract

Purpose:

To review the long-term duration of limb tumor complete remission (CR) and patient survival following therapeutic hyperthermic isolated limb perfusion (ILP) with cytotoxic drugs for melanoma.

Methods:

A retrospective case series of 124 ILPs performed in 111 patients.

Results:

There were 120 assessable ILPs. Patient staging (M.D. Anderson system) was stage II 11.7%, stage IIIA 44.2%, stage IIIAB 33.3%, and stage IV 10.8%. CR was initially attained after 83 ILPs (69.2%) and partial remission (PR) after 19 ILPs (15.8%). Limb CR was maintained in 28 (33.7%) of the 83 cases. Disease recurred in the perfused limb after an initial CR in the remaining 55 cases (median time to recurrence, 11 months); in 19 of these cases, the limb was disease-free at last follow-up after further locoregional treatment. A long-term CR was achieved, with or without further treatment, in 47 (56.6%) of the 83 cases in which an initial CR had occurred (mean follow-up, 97 months; median, 65 months). There was no significant difference in long-term local remission for stage IIIA and IIIAB patients. Five-year survival for those who had a partial or no response to ILP was 7%. Ten-year survival for those who had a long-term CR was 49%.

Conclusions:

ILP, with or without further locoregional treatment, achieved long-term control of recurrent and metastatic limb disease in 56.6% of cases in which an initial CR was achieved. A complete response to ILP was a positive prognostic indicator for survival, probably reflecting more favorable tumor biology in this subset of patients.

A 20-year review of isolated limb perfusions performed for melanoma at a single center showed a sustained, limb tumour complete remission (CR) rate of 39%. Ten-year survival for patients who achieved and remained in CR was 49%.

In 1997, the Sydney Melanoma Unit (SMU) reported the frequency and duration of remission after 114 consecutive isolated limb perfusions (ILPs) performed in the 10-year period to April 1994. It was found that 81 (73%) of the 111 assessable ILPs resulted in complete limb tumor remission.¹ However, when recurrences within the limb and further treatment were taken into consideration, a total of 56 of these 81 ILPs had a disease-free state in the limb after a median follow-up of 58 months.¹ The aims of this study were to determine the long-term duration of remission and survival following ILP in this patient group and those who underwent ILP following 1994.

METHODS

Data on remission and survival were collated for 120 of 124 consecutive, therapeutic ILPs performed between April 1984 and July 1997 using information from the SMU database. Two patients were excluded because they died of causes other than melanoma within 4 months of the ILP. Another patient was excluded because an early amputation of the perfused limb was performed and 1 patient was lost to follow-up. Current information on the health of patients discharged from regular follow-up at the SMU was obtained from their primary caring doctor or from the patient. The response rates were assessed according to WHO criteria.² CR was defined as the disappearance of all known limb disease; a partial response (PR) was defined as a 50% or more decrease in total tumor size; no change (NC) described a reduction of less than 50% in local tumor burden; and progressive disease (PD) was a 25% increase in the size of the lesions or the appearance of new lesions.

In total, 120 ILPs were performed in 108 patients. The technical details of the ILP procedures have been described previously elsewhere.^1,3,4 The duration of drug perfusion was 45 minutes for 72 ILPs and 60 minutes for 48 ILPs, and the maximum tissue temperatures were 40.5°C to 41.5°C. The toxicity of the ILPs is summarized in Table 1.⁵

TABLE 1. Postoperative Complications Following ILP According to the Wieberdink Classification⁵

Patients were selected for an ILP if they had bulky, inoperable disease or extensive limb disease that had developed after one or more attempts to clear it surgically. A local recurrence was defined as tumor deposits within 3 cm of the primary site, while an intransit metastasis was defined as tumor deposits between 3 cm of the primary site and the locoregional lymph node basin. To report the effect on patient outcome of the final remission state achieved, only the last ILP performed in the 9 patients who had repeat ILPs was included in the statistical analysis of survival. However, all 120 ILPs were considered in the analysis of remission rates, since excluding the unsuccessful ILPs would have produced a falsely greater response rate.

Statistical analyses were performed with SPSS 11 for Windows (2001, SPSS Inc., Chicago). Survival time was defined as months from ILP to death or to the date of last follow-up. Time to recurrence was defined as months from ILP to detection of new melanoma deposits within the limb. Cox regression analysis was used to evaluate the prognostic value of age, gender, stage, and final remission state for these 2 endpoints. Logistic regression was used to evaluate the prognostic value of age, gender, and stage on achievement of CR. After univariate analysis, all these factors were included in a multivariate model, with a stepwise backward algorithm to exclude factors with a prognostic significance of greater than 5%.

RESULTS

Patients

The patient characteristics are summarized in Table 2. Patients were staged at the time of ILP using the M.D. Anderson classification,⁶ because it more precisely categorizes limb disease than the standard AJCC/UICC classification and therefore is used widely by perfusionists. The M.D. Anderson classification is summarized in Table 3. Of the 108 patients involved in the study, 44 were male and 64 were female. Their median age was 61 years (range, 30–87 years). During the ILP, the drugs used were as follows: 20 patients received melphalan alone, 94 patients received melphalan and actinomycin-D, 1 patient had cisplatin and actinomycin-D, 1 had actinomycin-D alone, and 4 had cisplatin alone.

TABLE 2. Patient Characteristics

TABLE 3. M.D. Anderson Classification

Surgical control of local disease recurrences and in-transit metastases was attempted prior to the ILP in 86 patients. 54 patients had an average of 2.1 excisions of local recurrences, and 46 patients had an average of 2.0 excisions of in-transit metastases.

Patients with tumor recurrence following an initial complete response to ILP underwent further treatment determined by the burden of their disease, their comorbidities, and their personal preferences.

Initial Remission Rates

Of the 120 assessable ILPs, 83 (69.2%) achieved an initial complete remission, while 19 (15.8%) had an initial partial remission. The proportion of ILPs achieving an initial complete remission following ILP in relation to stage is presented in Fig. 1. The apparent increased likelihood of achieving an initial complete response in patients with a lower stage was not statistically significant.

FIGURE 1. Initial remission status as a function of staging.

Long-term Remission Rates

Complete remission was maintained until death or last follow-up in 28 (33.7%) of the 83 cases with no further treatment of local disease recurrence (median follow-up duration, 199 months; range, 8–226 months) (Fig. 2). Of the remaining 55 cases who had local recurrence following an initial CR, 41 had further treatment. The local therapy included isolated limb infusion (ILI) or ILP, surgical excision of local recurrences or in-transit disease, cryotherapy, and electrocautery. At last follow-up, 19 of these patients had achieved locoregional control after further treatment to the limb. A long-term disease-free state in the limb was achieved, with or without further treatment, in 47 (56.6%) of the 83 cases in which an initial complete remission occurred (mean follow-up, 97 months; median, 65 months; range, 8–244 months). Of these 47 patients, 37 had melphalan and actinomycin-D, 8 had melphalan alone, and 2 had cisplatin alone.

FIGURE 2. Flow chart describing the outcome of the 120 ILPs.

Seven patients underwent 2 ILPs, 1 patient had 3 ILPs, and 1 patient had 4 ILPs in the same limb for recalcitrant disease. Of these 9 patients who had repeat ILPs, 3 remained in complete limb remission at last follow-up.

At the last follow-up, 64.3% of stage II ILPs, 43.4% of stage IIIA ILPs, 37.5% of stage IIIAB ILPs, and 23.1% of stage IV ILPs were in locoregional remission (Fig. 3). Using multivariate analysis, there was a statistically significant difference between stages II and IV in achieving a long-term complete remission (P < 0.001). However, there was no statistically significant difference between stages IIIA and IIIAB.

FIGURE 3. Remission status at last follow-up as a function of staging.

Neither gender nor age was a significant prognostic marker for CR.

Recurrence and Limb Progression

The median time to recurrence for all ILPs was 9.3 months (mean, 17.7months; range, 1.7–183.3 months). Locoregional recurrences occurred in 55 ILPs who had an initial CR. Nineteen of these patients achieved locoregional control again after further treatment. Among these 19 patients who recurred but were in CR at last follow-up, the median time to recurrence was 15.3 months (mean, 34.2 months; range, 2.1–183.8 months). In those who had progressive limb disease at last follow-up, the median time to recurrence was 7.1 months (mean, 11.7 months; range, 1.7–66 months). Univariate analysis of the effects of age, gender, stage, and final remission state on limb progression revealed that only final remission state was a significant prognostic factor.

Survival

Of the 47 patients who had long-term complete locoregional remission, 24 have died at a median interval of 42.7 months (range, 8–150 months) after their last ILP. All but 2 patients died as a consequence of melanoma. Nineteen of the 23 patients who are alive and with CR have not developed systemic disease after a median interval of 175.4 months (mean, 155.1 months; range, 25.6–244.1 months). All 61 patients who were in partial remission or who had progression of their locoregional disease at last follow-up have died at a median interval from their last ILP of 16.1 month (mean, 21.5 months; range, 1–98 months).

The overall 5-year and 10-year survival rates for patients who had an initial CR were 40% (95% confidence interval [CI] 30%–53%) and 28% (CI, 19%–41%), respectively. The survival rates increased to 63% (CI, 50%–79%) and 49% (CI, 36%–66%), respectively, for patients who had remained in CR at last follow-up (Fig. 4). The 5-year survival in patients with progressive disease was 7% (CI, 2.5%–17%). Univariate analysis demonstrated that achieving and maintaining CR was a highly statistically significant prognostic indicator for survival (P < 0.0001) and females had a survival advantage over males (P < 0.01).

FIGURE 4. Survival according to long-term remission status.

When gender, stage (Fig. 5), and final remission state were compared in a multivariate analysis, complete remission at last follow-up continued to be highly significant in its association with survival, but the impact of gender and stage was reduced.

FIGURE 5. Survival according to M.D. Anderson stage.

DISCUSSION

In the 1997 report from the SMU on the frequency and duration of remission following ILP, the question of whether ILP could produce a prolonged and worthwhile remission or “cure” from locally extensive melanoma was raised.¹ After a median follow-up period of 177 months, 20 (18.5%) of the original 108 patients are alive and without any evidence of local or systemic disease. This confirmed that ILP is particularly indicated in those who did not have systemic disease and who may therefore survive for months or years with bulky and unsightly limb metastases that can ooze, smell, bleed, and cause pain and discomfort if ILP is not undertaken.

The reported 5-year survival rate following ILP for stage II disease is 57% to 80%, for stage IIIA disease 35% to 70%, for stage IIIAB disease 23% to 40%, and for stage IV 8% to 10%.^7–11 Reports documenting 10-year survival are uncommon. Previous studies report that the 10-year survival rate following ILP for stage II disease is 58% to 63%, for stage IIIA disease 28% to 50%, and for stage IIIAB disease 16% to 34%.^7,8 Our results were within these ranges. The statistical variations between studies probably reflect differences in patient cohorts and ILP techniques.

It is important to understand the implications of a statistically significant association between response to ILP and survival. In our study and others, the achievement of complete remission was a highly significant prognostic variable with respect to survival.^12,13 ILP cannot be expected to alter the occurrence or progression of systemic disease because it is a locoregional treatment modality. Indeed, it is based on the principle that the limb is isolated from the systemic circulation. The suggestion that ILP might arrest the lymphatic and hematogenous dissemination of melanoma seems implausible, since there was no significant survival difference demonstrated between patients with M.D. Anderson stage IIIA and stage IIIAB disease. This implies that the subset of patients who are alive and disease free must be presumed to have less aggressive tumor biology, but this could not be confirmed with any biologic correlates such as tumor markers. This is further supported by previous studies that showed a 15% to 30% 5-year survival for patients with locally advanced melanoma treated by amputation.¹⁴

There have been 2 randomized trials examining the effect of adjuvant ILP on survival. The first reported a survival advantage but has been criticized for its very small sample size (n = 107) and unusually poor outcome in the control arm.¹⁵ In the larger, multicenter WHO/EORTC trial, a reduction in locoregional recurrence rates after adjuvant ILP did not translate into an improvement in overall survival, and there was no impact on the prevention of distant metastases.¹⁶

In our study, the markers of tumor aggressiveness were disease stage at the time of ILP and time to recurrence. Patients with stage IV disease were significantly less likely to have a prolonged CR when compared with stage II patients. There was a trend for better survival and response rates in patients who had histologically negative lymph nodes, but this finding was not statistically significant. While some groups have found significant remission or survival benefits from the absence of nodal metastases,^11,13 others have not.^12,17 This inconsistency between groups can again be attributed to the great variation between patient cohorts and ILP techniques. With the exception of M.D. Anderson stage II versus IV, there were no patient characteristics identified prior to the ILPs that might select out those who will have a better response to locoregional chemotherapy.

Interestingly, in a similar review of 207 isolated limb infusions (ILIs) performed in our unit, there was an obvious decrease in survival between stages IIIA and IIIAB disease. Other significant prognostic markers for survival included fewer tumor deposits, a melphalan concentration at 5 minutes of over 67 μg/mL and final limb subcutaneous temperature over 37.8°C. ILIs and ILPs performed at the SMU show similar remission rates despite significant differences in their technique.^1,18–20 ILP involves open surgical cannulation of iliac, femoral, or axillary vessels, with the application of a tourniquet at the root of the limb.³ ILI is performed through cannulas inserted percutaneously under radiologic guidance into the contralateral groin and passed into the popliteal or brachial vessels, with the tourniquet applied at the upper limit of macroscopic limb disease, and not necessarily at the root of the limb.¹⁹ ILP therefore includes the regional lymph nodes in its treatment field in most patients, whereas ILI does not. This suggests that ILP may have achieved similar survival and response rates in IIIA and IIIAB patients in our study because it sterilized the regional lymph node basin. This is supported by the results of the WHO/EORTC prophylactic ILP trial, involving patients with melanomas between 1.5 and 2.99 mm in thickness, in which there was a reduction in regional lymph node metastases in those treated by ILP compared with those in the control group.¹⁶

It has been proposed that an indicator of tumor-host interaction is the time to disease recurrence.²¹ The median time to first limb recurrence was 15.3 months in patients who had a sustained complete remission and only 7.1 months in patients who had progressive limb disease. Time to recurrence is thus a practical and clinically applicable indicator of patient prognosis. In this study, the time to recurrence was strongly associated with improved outcome and was the only patient variable that might reliably predict long-term response to the ILP. There was no statistical difference between stage IIIA and stage IIIAB patients in time to first limb recurrence after ILP. This is perhaps explained by a reduction in regional lymph node recurrences, as the regional lymph node basin was resected at the time of or prior to the ILP in most patients.

If the proposition that patients with delayed recurrences have less aggressive tumor biology is correct, then it is likely that they will be more amenable to disease control with simple local measures such as surgical excision, electrocautery, and CO₂ laser ablation. Sustained complete remission was achieved in our series in 19 of the 55 patients who had a recurrence after an initial complete response to ILP. One of these patients eventually achieved and maintained complete limb remission after self-treatment of his limb recurrences with solid CO₂ over a period of many years. These observations suggest that, in some patients at least, both limb and systemic recurrences are due to dissemination of melanoma cells or cell clusters prior to removal of the primary melanoma.

The remission rates achieved by ILP have not been compared in a randomized trial to remission rates after other local treatments, such as cautery, surgery, laser, or radiotherapy. Although ILP does not stand alone as a treatment of extensive limb disease, it is clearly superior to other local techniques in treating the whole limb, as opposed to a single area of macroscopic disease.²² However, this also implies that the whole limb is exposed to the potentially deleterious effects of cytotoxic drugs and hyperthermia.

ILP has been shown previously to increase the limb recurrence-free interval and decrease the number of lesions per recurrence episode compared with surgical excision alone.²³ This indicates that ILP can target microscopic disease. It has been suggested that the treatment is useful in those whose limb recurrence-free intervals are shortening over time.²⁴ However, ILP is likely to mediate its effects by cytotoxic or hyperthermia-induced damage to tumor blood supply and will have only a limited ability to target micrometastases that do not have a mature vascularity.¹

Of the 9 patients who had repeat ILPs in our study, 3 successfully achieved a long-term limb CR. While these numbers are too small to assess statistically, they further suggest that ILP is less effective against tumors lacking a well-developed vascular supply rather than because of tumor tolerance to the chemotherapy agents. The benefits of repeat ILP for melanoma recurrences have been shown in other studies.^24,25

CONCLUSION

An initial complete remission of limb disease was achieved in 69% of 120 ILPs performed in our unit. In long-term follow-up, this was maintained in 40% of ILPs, with 18% of patients alive and completely free of metastatic disease at last follow-up. ILP is thus beneficial as a palliative procedure in those with locally extensive melanoma of a limb, both by achieving CR and prolonged PR, and by delaying the onset of new recurrences. In a subset of patients with less aggressive tumor biology, it is also associated with prolonged disease-free survival, which might perhaps be regarded as “cure,” but to determine whether this is due solely to the ILP or to the inherent tumor biology would require confirmation in a randomized controlled study.

ACKNOWLEDGMENTS

The authors thank Marjorie Colman, BSc, Database Manager of the Sydney Melanoma Unit, for her assistance in the statistical analysis of our data.