Alternative Therapies: Regional Chemotherapy for Malignant Melanoma
Danielle M. Bello and Charlotte E. Ariyan
Half of all patients with cutaneous melanoma have their primary lesion located on the extremity; of those with a primary melanoma of Breslow depth ≥1.5 mm, 15% will develop a local recurrence or in-transit recurrence (1,2). These in-transit melanomas are metastatic foci that develop between the primary lesion and the draining lymph node basin due to lymphatic permeation. They present as subcutaneous or cutaneous papules or nodules, which can be blue-grey, purple, or red in appearance and are often intralymphatic (Figure 12.1). In-transit disease has significant morbidity associated with it in the form of pain, ulceration, bleeding, and infection. Therapeutic management of extremity disease is an important issue as median survival after diagnosis of in-transit or locally recurrent disease, classified as stage IIIB or IIIC depending on regional nodal involvement, is approximately 20 months (3–5).
Regional therapy, both isolated limb perfusion (ILP) and isolated limb infusion (ILI), were developed with the goal of controlling extremity disease. ILP, and its less invasive counterpart ILI, were established as an alternative to amputation in patients with recurrent cancer of the extremity. The concept was that vascular isolation of the limb would allow delivery of higher—and potentially more effective—doses of chemotherapy to the diseased limb than could be achieved with systemic therapy. While radical surgical techniques to eliminate extremity recurrence are rarely used, in a highly selected group of patients undergoing amputation, there was a reported 42% 5-year survival rate (6).
This chapter will highlight the historical basis as well as the significant contributions to the evolution of regional chemotherapy of the extremity that ultimately culminated in the randomized controlled trials that today form our treatment paradigms. Recently, isolated regional therapy in combination with other types of systemic therapeutic agents such as chemotherapy and immunotherapy have emerged to further change the landscape of regional management of melanoma and offer a glimpse of not just palliation but improved survival outcomes. Trials investigating these combinations are underway and the preliminary results are promising.
THE HISTORY OF REGIONAL CHEMOTHERAPY
Isolated Limb Perfusion
In 1958, Creech and Krementz (7) first described the ILP procedure, which was performed at Charity Hospital in New Orleans, Louisiana. ILP is performed under general anesthesia with regional lymphadenectomy being performed simultaneously. The extremity is isolated from the systemic circulation using a tourniquet and open surgical cannulation of the femoral vessels. Next, an extracorporeal circuit introduces a high-flow, hyperoxic perfusate into the limb, thereby achieving adequate tissue perfusion pressures and allowing relatively long treatment duration. Melphalan, the nitrogen mustard alkylating agent, was the first chemotherapeutic agent used based on in vivo data suggesting significant antitumor activity in mice (8,9). Melphalan was felt to be an ideal agent for regional chemotherapy due to its short half-life and its predictable linear dose-response curve in relation to its cytotoxicity (7,10). Following the perfusion, the circuit is washed out with normal saline, the cannulae are withdrawn, and the vessels are repaired (7,11,12).
The field of regional perfusion was further advanced by Cavaliere et al. (13) in 1967 with the findings that hyperthermia could increase the therapeutic efficacy of ILP. They investigated the effects of hyperthermic perfusion without chemotherapy in humans with recurrent extremity tumors. The study included 22 patients with various extremity malignancies including squamous cell carcinoma, melanoma, and sarcoma who were treated with hyperthermic perfusate. The duration of hyperthermia ranged from 50 minutes to more than 6 hours and response to therapy was documented by tumor biopsy both pre- and posttreatment. Complete response (CR), described as complete disappearance of the tumor, was demonstrated in 10 of 22 patients (45%). Morbidity and mortality, however, was significant as six patients (27%) died in the immediate postoperative period. Of the 16 survivors, 12 patients demonstrated durable responses and were alive without evidence of disease at 3 to 28 months of follow-up (13). Of all the tumor subtypes, melanoma was noted to be the most responsive to hyperthermic therapy.
Given these findings of the deleterious effects of temperature on tumor cells, Stehlin et al. (14) chose to combine regional chemotherapy with heated perfusion for treatment of extremity malignancy. He modified his protocol by increasing both the temperature of the perfusate (to 42°C) and the perfusion time from 45 minutes to 2 hours. Of the 50 patients undergoing both therapeutic and adjuvant ILP, patients with melanoma received melphalan, and sarcoma patients received melphalan and dactinomycin. Of the 37 melanoma patients, only 12 were performed for macroscopic disease. Of these, 10 of the 12 patients (83%) had a response defined as “pronounced regression” which was sustained for 3 months or more. This response represented an improvement over the conventional normothermic chemotherapy, but whether these results represented a CR remained unclear. Morbidity and mortality of the combined approach was also high with two deaths reported in addition to significant postoperative edema (70%), hemoglobinuria (20%), and bleeding (18%) complications (14).
Isolated Limb Infusion
In 1993, John Thompson of the Melanoma Institute of Australia (formerly the Sydney Melanoma Unit) first described ILI, a new form of regional chemotherapy for extremity tumors (15). Rather than the open vasculature exposure required in ILP, ILI is performed by percutaneously placing angiographic catheters into the contralateral groin using the Seldinger technique under radiographic guidance to gain access to the femoral vessels in the affected limb. This eliminates the need for open surgical cannulation of the vessels, thereby making it a less invasive approach to treating regional disease. After systemic heparinization, an Esmarque tourniquet or blood pressure cuff is positioned on the proximal thigh of the diseased limb, above the level of the catheter tips (which are located above the knee), and inflated to 250 to 350 mmHg. Once the extremity has been effectively isolated from the systemic circulation, melphalan and dactinomycin are infused into the arterial catheter, and then the limb infusate (chemotherapy and limb blood volume) is circulated through a blood warmer using manual pressure. The limb is warmed during the procedure to at least 38°C to 39°C. During the procedure, the perfusion is not oxygenated as in ILP and, therefore, the duration is limited to approximately 30 minutes. The hypoxic and ischemic conditions are thought to increase the efficacy of cytotoxic agents but also can contribute to soft tissue damage. After completion of the ILI, a crystalloid solution is infused, and venous outflow is discarded to flush the extremity of chemotherapy before the procedure is terminated.
ILI offers several distinct advantages over ILP, particularly with regard to ease of performance. No perfusionist is required for ILI, as the perfusion is performed manually during the procedure. The infusion is low flow, normothermic or slightly hyperthermic, hypoxic, and acidic, resulting in less leakage into the systemic circulation as well as potentiation of the chemotherapeutic effect. No pump priming is required, the duration of the procedure is short, and patients who may not tolerate ILP due to comorbid vascular conditions can be treated with ILI. In addition, due to its less invasive nature, the procedure can be repeated with more ease than ILP. Potential disadvantages to ILI include the difficulty in treating more proximal lesions of the extremity due to the more distal position of the catheter tips, and the need to perform a separate incision if a simultaneous nodal dissection is required.
Thompson’s group reported the first clinical experience with ILI, examining patients with recurrent melanoma of the extremity or recurrent sarcoma. Of the 82 patients who had a minimum of 6 months follow-up, 39% had a CR and 52% had a partial response (PR), with a median follow-up of 16 months (16). These data were supported by Brady et al. (17), who reported a prospective phase II trial of ILI with melphalan and dactinomycin in patients with stage IIIB or IIIC melanoma or unresectable soft tissue sarcoma. In this study, at 1 year 23% of patients had a CR with a median duration of 12 months, and 27% had a PR with a median duration of 11 months (17). Thus, ILI emerged as another legitimate beneficial therapeutic technique in regional disease management.
DRUGS OF REGIONAL THERAPY
Melphalan is the most commonly employed agent used in both ILI and ILP and was first described in Creech’s original perfusion (7). Melphalan is a derivative of the amino acid phenylalanine, which is known to play a vital role in melanin synthesis and is preferentially taken up by melanocytes (18). Thus, melphalan has been hypothesized to produce selective toxicity in melanin-containing melanoma tumor cells. Melphalan functions as an alkylating agent, inducing its cytotoxic effect by alkylating DNA bases and crosslinking DNA strands (19). Systemically, the therapeutic dose of melphalan exceeds the maximally tolerated or toxic dose, limiting its utility in this capacity. The effective dose, however, is easily achieved regionally. In early studies, the dose was calculated based on bodyweight rather than on the affected limb size. This method has now been replaced by calculation of limb volume, tailoring dosage to limb mass and limiting the risk of regional toxicity (20). Traditionally, melphalan was used as a single agent; however, more recently melphalan has been used in combination with actinomycin in ILI, an antibiotic containing a cyclic polypeptide that binds to DNA and inhibits RNA synthesis, as well as tumor necrosis factor alpha (TNF-α) in ILP (21).
Other Chemotherapeutic Agents
Other chemotherapies have been evaluated for efficacy in regional therapy in an effort to improve on the results seen with melphalan. These include dimethyltriazeno imidazole carboxamide (dacarbazine or DTIC), doxorubicin, cisplatin, carboplatin, thiotepa, and more recently temozolomide (TMZ). In 1978, Ariyan et al. (22) were the first to use an agent other than melphalan as a perfusate. This agent, DTIC, demonstrated a low toxicity and risk profile with notable clinical benefit observed in the 24 patients treated. However, despite its low toxicity as confirmed by others, response rates were also lower than seen with melphalan, with 12% of patients having a CR and 29% experiencing a PR to ILP with DTIC (23). In a retrospective series of patients treated with therapeutic ILP with primarily DTIC or carboplatin, 26% of patients had a sustained CR at a median follow-up of 58 months (24).
Cisplatin is another agent that has been used in ILP. Cisplatin, an inhibitor of DNA synthesis with significant potential for nephrotoxic and neurotoxic side effects, showed preclinical and clinical potential as a regional chemotherapeutic agent. When used in combination with ILP, cisplatin demonstrated CR rates of 11% at 3 years (25). However, its use was limited by significant limb-threatening toxicity; 10% required an amputation, and 20% had severe tissue toxicity (26,27). Unfortunately, there is a paucity of phase I or phase II trials to support the role of these other agents and melphalan remains the most commonly utilized treatment.
Recently TMZ, another DNA alkylating agent similar to melphalan, showed early potential in preclinical animal experiments compared with melphalan (28). Additionally, there is new evidence to suggest TMZ may have activity in melphalan-resistant melanoma tumors (29). A recently completed phase I multicenter trial examining the maximally tolerated intra-arterial dose of ILI with TMZ in patients with regionally advanced melanoma of the extremity who had previously failed melphalan ILI showed an overall response rate of 16% (10.5% CR, 5% PR) at 3 months postinfusion. Although the numbers are small, this study did show that some patients who are unresponsive to ILI with melphalan could achieve a CR to ILI with TMZ (30). One hypothesis for the low response rates was that acidotic conditions of ILI prevented TMZ from being converted to its active metabolite, suggesting a role for TMZ with ILP; however, further investigation is needed.
Immune Agents in Regional Therapy
TNF-α is a cytokine discovered in the serum of animals during investigations into the mechanisms of Bacillus Calmette-Guerin (BCG). Injections of mice with BCG and endotoxin resulted in release of a factor that induced necrosis in a murine model of sarcoma within 24 hours, with approximately 20% of the tumors disappearing completely (28,31,32). Further studies demonstrated synergy and enhanced tumoricidal effect of TNF-α when combined with interferon (IFN) α, β, or γ in murine models (33). Correlative studies in human trials with systemic TNF-α therapy were limited by its severe side effects, similar to those seen in septic shock.
In ILP, however, the toxicity of systemic TNF-α exposure could be abrogated by vascular isolation of the limb, making it an appealing agent. TNF-α is thought to increase the efficacy of ILP by two mechanisms. First, it causes vasodilation of the tumor vasculature, allowing high concentrations of cytotoxic agents such as melphalan to penetrate the tumor (34,35). Second, at 1 to 2 days postperfusion it causes coagulation of immature vasculature, locking in the cytotoxic agents and rendering the tumor ischemic (36).
Early human trials investigating the role of TNF-α alone with hyperthermic ILP showed low and limited response rates (37). Despite the disappointing results of TNF-α alone, further investigations prompted a phase II trial combining high-dose TNF-α, IFN-γ, and melphalan in ILP. IFN was added because of potential antitumor synergism (33,38). In this trial, all 19 patients had significant extremity in-transit disease with a median of 10 lesions per patient. The treatment was universally toxic; all patients had a systemic inflammatory response requiring dopamine prophylaxis. Two patients required an amputation due to significant limb toxicity. At a mean follow-up of 11 months, however, 89% of patients had a CR and 11% had a PR to the combination treatment (39). This initial report resulted in tremendous interest in the addition of TNF-α to ILP, resulting in many single-institution, nonrandomized trials. The role of IFN as a component of the ILP was also separately addressed in a phase II trial; however, no significant benefit was observed with the addition of IFN to TNF-α and melphalan (CR 69% vs. 78%; not significant), and its use has subsequently waned (40).
The 2006 American College of Surgeons Oncology Group (ACOSOG) Z0020 trial compared ILP and melphalan to melphalan and TNF-α in the treatment of patients with in transit melanoma (41). One hundred and three patients with recurrent melanoma of the extremity were enrolled; however, the trial was stopped early due to increased toxicity in the combination arm without evidence of improvement over melphalan alone (25% CR with melphalan vs. 26% CR with melphalan and TNF-α; p = .89). Overall, the rate of complications was not significantly higher in patients receiving combination therapy (33% had adverse events of at least grade III with melphalan alone vs. 37% with melphalan and TNF). However, the rate of grade IV adverse events was significantly increased in the combination group (11% vs. 1% in melphalan alone group). In addition, there was one amputation secondary to disease progression in the melphalan group, whereas the combination therapy resulted in two patients with amputations necessary due to toxicity. This trial has been criticized for the early time point for assessing response (3 months) and for the lower response rates observed with TNF-α compared with prior trials (42). Unfortunately, this negative experience has abrogated the use of TNF in the United States. Further trials in Europe have demonstrated the safety and efficacy of TNF, which is used with enhanced response rates (43).
Some degree of soft tissue injury is usually seen following regional therapy, most likely due to the high concentration of toxic agents delivered to the extremity. The most commonly used grading system to evaluate local complications of regional therapy is that described by Wieberdink et al. (20) (see Table 12.1). Grading ranges from I (in which there is no evidence of adverse effect) to grade V (representing severe limb-threatening reactions). Grades I and II are mild, and while grade III may lead to some impairment in limb function, the majority resolve without incident. Grade IV to V reactions are more severe and can lead to long-term morbidity, including amputation.
Overall, although mild grade I to II toxicity has been cited as high as 85% in larger series, grade V toxicity affects less than 1% of all patients (44). The range of toxicity is similar for ILI and ILP, but ILP carries a higher risk of more severe toxicity. Following ILI with melphalan and actinomycin, grade III toxicity was observed in 10% to 53% of patients, whereas grade IV reactions were seen in 2% to 15% (17,45–53). Grade V reactions were rare with only one amputation reported in these studies (45). The proportion of grades III and IV reactions following ILP with melphalan and TNF-α are similar to those with ILI with rates of 4% to 26% and 0% to 20%, respectively (54–62). The risk of compartment syndrome and amputation is higher in ILP although low overall (0%–2%) (54–61,63).
Mild edema or erythema
Considerable edema or erythema with blistering; some functional impairment
Extensive epidermolysis or damage to the deep tissues; definite functional impairment; threatened or manifested compartment syndrome
Major tissue damage necessitating amputation
ILI, isolated limb infusion; ILP, isolated limb perfusion.
Multiple strategies have been developed to reduce the risk of toxicity associated with regional therapy. One previously discussed technique was adjusting dosage to limb volume. Additionally, for isolated proximal lesions on the extremity, the distal-most aspect of the limb—the hand and foot—can be excluded from exposure to chemotherapy by using an additional tourniquet. After surgery, limb elevation may help improve edema. Finally, close clinical examination and serial creatine phosphokinase (CPK) monitoring can also raise concern for significant soft tissue injury and help to diagnose compartment syndrome early if present.
EFFICACY AND OUTCOMES
Although both ILP and ILI are effective, there have not been any randomized trials comparing the two techniques directly to one another. Direct analysis is limited to extrapolated comparisons from different studies with varied patient populations, nonuniform protocols, and levels of experience. Methods used to assess responses include clinical evaluation of disease burden, serial radiological measurements, or, in the case of induction regional chemotherapy, percentage of tumor necrosis of resected specimen.
The best response rates for regional chemotherapy in melanoma have been seen in in-transit melanoma metastases. However, all evidence to date supports the role for ILP and ILI as mainly palliative rather than curative. In 1998, Koops et al. (2) performed a randomized controlled trial of 832 patients over a 10-year period by a collaborative effort with the European Organization for Research and Treatment of Cancer (EORTC), the World Health Organization (WHO), and the North American Perfusion Group Southwest Oncology Group (SWOG). Patients with extremity melanoma ≥1.5 mm thick, without evidence of distant disease, in-transit disease, or nodal disease, were randomized to wide local excision (n = 422) or wide local excision with ILP (n = 430). Limb perfusion was performed with melphalan and hyperthermic ILP for 1 hour. While there was a decrease in local recurrence after ILP, after a median follow-up of 6.4 years, there was no difference in survival or time to development of distant metastasis. Similar findings were reported by Hafstrom et al. (64), who examined 69 patients at the time of their initial in-transit recurrence, randomizing patients to wide local excision (WLE) or WLE and regional adjuvant chemotherapy. As a result of these studies, ILP was no longer recommended as an adjuvant treatment for patients with extremity melanoma, theorizing that local management of the disease had not proven to influence the development of distant metastases. Instead, regional chemotherapy became reserved for bulky extremity disease in order to palliate symptoms.
Outcomes following ILP have been variable, reflecting diverse patient populations and treatment strategies; however, response to therapy is typically seen within 6 to 12 weeks after perfusion (Figure 12.2). Although single center studies have described CR rates of 26% to 91%, PR rates of 9% to 50%, and overall response rates of 69% to 100% (Table 12.2) (10,39,40,56,62,63,65–71), the multicenter ACOSOG Z0020 study (41) reported a lower CR rate. As previously mentioned, the ACOSOG Z0020 trial was a prospective, multicenter randomized trial comparing outcomes with melphalan to melphalan plus TNF-α in ILP. In this study, in which there was accurate measurement of pre- and posttreatment disease, OR rates of 64% and CR rates of 25% were demonstrated with hyperthermic ILP for patients with melanoma; there was no significant difference in outcomes between the two arms (41). While some data suggest CRs are associated with short-term disease control and are more durable compared with PRs (40,47,48,71), consistent cumulative results indicate responses are not durable with 50% to 60% recurrence rates within the first year and overall 5-year survival of 30% to 40% (72).
Compared with ILP, ILI outcomes have been somewhat inferior. The overall response rates following ILI range from 43% to 100% with CR rates of 23% to 44% and PR rates of 27% to 56% (see Table 12.3) (17,45–49,52,53,71,73–75). In examining recurrence patterns, ILI has been associated with a decreased time to first recurrence (8 months vs. 23 months) and an overall higher probability of recurrence (85% vs. 65%) compared with ILP (76). Despite these differences, no significant differences in overall survival (OS) have been identified between the two treatment groups.