Factors predicting toxicity and response following isolated limb infusion for melanoma: An international multi-centre study
Timothy J. Kenyon-Smith, Hidde M. Kroon, John T. Miura, J. Teras, Georgia M. Beasley, Dean Mullen, Norma E. Farrow, Paul J. Mosca, Michael C. Lowe, Clara R. Farley, Aishwarya Potdar, Hala Daou, James Sun, Jeffrey M. Farma, Michael A. Henderson, David Speakman, Jonathan Serpell, Keith A. Delman, B. Mark Smithers, Andrew Barbour, Brendon J. Coventry, Douglas S. Tyler, Jonathan S. Zager, John F. Thompson
Abstract
Introduction: Isolated limb infusion (ILI) is a minimally-invasive procedure for delivering high-dose regional chemotherapy to treat melanoma in-transit metastases confined to a limb. The aim of this international multi-centre study was to identify predictive factors for toxicity and response.
Methods: Data of 687 patients who underwent a first ILI for melanoma in-transit metastases confined to the limb between 1992-2018 were collected at five Australian and four US tertiary referral centres.
Results: After ILI, predictive factors for increased limb toxicity (Wieberdink grade III/IV limb toxicity, n=192, 27.9%) were: female gender, younger age, procedures performed before 2005, lower limb procedures, higher melphalan dose, longer drug circulation and ischemia times, and increased tissue hypoxia. No patient experienced grade V toxicity (necessitating amputation). A complete response (n=199, 28.9%) was associated with a lower stage of disease, lower burden of disease (BOD) and thinner Breslow thickness of the primary melanoma. Additionally, an overall response (combined complete and partial response, n=441, 64.1%) was associated with female gender, Australian centres, procedures performed before 2005, lower limb procedures and lower actinomycin-D doses. On multivariate analysis, higher melphalan dose remained a predictive factor for toxicity, while lower stage of disease and lower BOD remained predictive factors for overall response. Conclusion: ILI is safe and effective to treat melanoma in-transit metastases. Predictive factors for toxicity and response identified in this study will allow improved patient selection and optimization of intra-operative parameters to increase response rates, while keeping toxicity low.
Keywords: melanoma, regional chemotherapy, isolated limb infusion, loco-regionally metastatic disease, multi-centre, toxicity, response.
Introduction
Metastatic melanoma is generally associated with a poor prognosis. However, some patients develop in-transit metastases confined to a limb without evidence of systemic disease. (1) These patients constitute a specific group with particular treatment challenges. The limb lesions are often too numerous or bulky for surgical excision, while the efficacy and appropriateness of systemic therapies in this subset of patients is not known as it has not yet been studied specifically, and serious systemic side-effects can occur when they are used (2-4).
In this circumstance, delivering regional chemotherapy by isolated limb infusion (ILI) has been shown to be safe with negligible systemic side-effects and mild to moderate limb toxicity in most cases, while achieving satisfactory and durable responses. (5,6) Historically, however, ILI reports are mostly single-centre experiences or combined analyses from one country (7-9). Furthermore, to understand the current role of ILI as treatment alternative for melanoma in-transit metastases to the emerging systemic therapies, it would help clinicians to confirm current and identify new predictive factors for toxicity and response, allowing for better patient selection and further optimization of intra-operative factors. Therefore, the aim of this study was to determine factors predicting toxicity and response following ILI based on an international, multi-centre experience.
Patients and Methods
Post approval by each center’s ethics research committee or institutional review board, a multi-centre cohort study was conducted using prospectively collected data of patients undergoing ILI between 1992 and 2018. All included patients had melanoma in-transit metastases (AJCC 7th edition stage IIIB/IIIC) confined to a limb and underwent a first ILI at five Australian and four US participating centres (1).
Patients who had undergone treatment for in-transit metastases (such as isolated limb perfusion (ILP), intra-lesional therapy, surgical excision or systemic therapy) prior to the first ILI were considered for the study. All institutions had established expertise in performing ILI (5). The ILI procedures were performed as described previously (10), using a combination of melphalan (7.5mg/L for lower extremities and 10mg/L for upper extremities) and actinomycin-D (100µg/L). Drug dosages were based on limb-volume measurements. For large limb volumes, the maximum melphalan dosage was restricted to 100mg for lower limb ILI, and 50mg for upper limb ILI. The melphalan dose was corrected for ideal body weight in the US centres, as previously described (11).
In patients with metastatic disease in their inguinal or axillary lymph nodes, a regional lymphadenectomy was undertaken following the ILI procedure under the same general anesthetic, after heparin reversal. Following the ILI procedure, patients were closely monitored with regular physical examination and measurement of serum creatine phosphokinase (CPK) levels daily. The Wieberdink scale was used to assess limb toxicity (12). World Health Organization (WHO) criteria for reporting results of cancer treatment were used by the Australian centres, in which the best response is captured after two observations >4 weeks apart (13). At the US centres, response was determined at 3 months after ILI according to the Response Evaluation Criteria in Solid Tumors (RECIST) guidelines for cutaneous lesions (14). Additionally, according to the WHO criteria a partial response (PR) is determined as a ≥50% decrease in total tumour size, whereas the RECIST guidelines state a PR as a ≥30% decrease. For the present study, it was our aim to identify predictive factors for the previously reported toxicity and response after ILI. (5)
The following clinico-pathological data were recorded; age, gender, country, treatment period, AJCC stage, Breslow thickness of the primary melanoma, involved extremity (upper or lower), burden of disease (BOD: low ≤10 lesions and no lesion >2cm, high >10 lesions or any lesion >2.0cm) (15). Peri-operative data collected were: limb volume, melphalan dose, actinomycin-D dose, drug circulation time, tourniquet time, initial and 30-minute intramuscular limb temperatures, and 30-minute PaO2, pH and base excess. Postoperative data that were collected included: CPK peak and on which postoperative day, Wieberdink limb toxicity grade, length of stay (LOS) in hospital and response to treatment.
Continuous variables are presented as medians with interquartile ranges (IQR), and categorical variables as frequencies and percentages. Where median values of comparison groups were identical, mean and standard deviation (SD) are also reported. For the univariate analyses, the Chi2 test was used to compare frequency distributions and the Mann–Whitney U test was used for non-parametric variables. Significance was based on pvalues <0.05 and 95% confidence intervals (95%CI). Multivariate analysis of toxicity and response was performed using multi-nominal logistic regression. Alpha was set at 0.05. Statistical analyses were performed using IBM SPSS statistical software version 25.0.0 (SPSS Inc., Chicago, IL, USA).
Results
Patient and tumour characteristics for the complete cohort are displayed in Table 1. For the 687 patients who were included, the median age was 71 years (IQR=17) and 412 patients (60%) were female. The majority of patients underwent ILI in Australia (60.5%), with more cases performed in 2005 or after (62.7%). 376 of the patients (56%) had stage IIIB ( in-transit metastases without regional lymph nodal involvement) melanoma and most (610) underwent lower limb procedures (89%). Table 2 lists the perioperative data for the complete cohort, and the toxicity and clinical outcomes as reported earlier (5).
Toxicity
Female patients experienced more Wieberdink grade III/IV limb toxicity than males (p=0.002, odds ratio 1.75, 95%CI 1.22-2.50, Table 3a). Patients who underwent ILI before 2005 experienced greater toxicity relative to post 2005 procedures (p=0.019, odds ratio 1.49, 95%CI 1.06-2.13). Also, patients undergoing lower limb procedures had more grade III/IV limb toxicity compared to upper limb procedures (p=0.02, odds ratio 2.04, 95%CI 1.103.85). Compared to patients with grade I/II limb toxicity, those with grade III/IV limb toxicity were administered higher melphalan dosages (median 45 vs. 42 mg, p=0.001), had longer drug circulation times (mean 26 vs. 25 minutes, p=0.012), longer limb ischemia times (median 58 vs. 53 minutes, p=0.002), and lower tissue oxygen levels at 30 minutes (median 10 vs. 11mmHg, p=0.01). After ILI, patients with grade III/IV limb toxicity had significantly higher serum CPK levels (median 1608 vs. 339 IU/L, p=0.001), a later CPK peak (median 5 vs. 4 days, p=0.001) and a longer LOS (median 9 vs. 6 days, p=0.001). Table 3b lists the analysis of grade IV limb toxicity (n=27) compared to grade I/II/III toxicity (n=652). Those with grade IV toxicity all underwent lower limb procedures (100% vs. 88.3% for grade I/II/III, p=0.01, odds ratio 1.04, 95%CI 1.03-1.06), were more likely to have received treatment in a US centre (p=0.032, odds ratio 2.31, 95%CI 1.05-5.06), were younger (median 58 vs. 71 years, p=0.001), had greater limb volumes (median 7 vs. 6 liters, p=0.001), had higher doses of melphalan administered (median 50 vs. 40mg, p=0.001), had longer circulation times (median 31 vs. 30 minutes, p=0.005) and longer limb ischemia times (median 59 vs. 54 minutes, p=0.001). Postoperatively, patients who experienced grade IV limb toxicity had a higher CPK peak (median 4601 vs. 711 IU/L for grade I/II/III limb toxicities, p=0.001), a later CPK peak (median 5 vs. 4 days, p=0.001) and a longer LOS (median 11 vs. 7 days, p=0.001). In the multivariate analysis, a high melphalan dose remained a significant factor for grade III/IV limb toxicity (p=0.03, odds ratio 1.05, 95%CI 1.01-1.10. No factor remained significant for grade IV limb toxicity in the multivariate analysis.
Response
Predictive factors for complete response (CR) are summarized in Table 4a. Patients with a CR following ILI had a lower stage of disease (p=0.005, odds ratio 1.62, 95%CI 1.15-2.28), a lower BOD (p=0.001, odds ratio 1.85, 95%CI 1.32-2.56) and a thinner primary melanoma (median 2.25 vs. 3.00 mm, p=0.001). No intra-operative factor was predictive for CR. Table 4b summaries predictive factors for overall response (OR, defined as the combined complete [CR] and partial response [PR]). For OR, predictive patient factors were: female gender (p=0.03, odds ratio 1.43, 95%CI 1.03-1.96), lower limb procedures (p=0.005, odds ratio 1.97, 95%CI 1.22-3.20), lower stages of disease (p=0.001, odds ratio 2.16, 95%CI 1.56-2.97), lower BOD (p=0.05, odds ratio 1.37, 95%CI 1.01-1.89), Australian procedures (p=0.001, odds ratio 2.4, 95%CI 1.73 – 3.32), ILI performed before 2005 (p=0.001, odds ratio 1.79, 95%CI 1.27 – 2.51) and a thinner primary melanoma (median 2.50mm vs. 3.00 mm, p=0.002). Treatment-related factors associated with OR were a lower actinomycin-D dose (median 450 vs. 500μg, p=0.001) and a later postoperative CPK peak (mean 4.3 vs. 3.8 days, p=0.001). Multivariate analysis revealed no significant factors for CR. In the multivariate analysis of OR, stage IIIB disease (p=0.007, odds ratio 2.88, 95%CI 1.32 - 6.33) and a low BOD (p=0.04, odds ratio 2.22, 95%CI 1.03 - 4.76) remained significant predictive factors.
Discussion
This international multi-centre study investigating predictive factors for toxicity and response in 687 patients undergoing ILI for melanoma in-transit metastases confined to a limb has previously shown that limb toxicity following ILI is mild to moderate (Grade I/II/III in 94.9%), without any patient requiring a toxicity-related amputation (grade V) (5). Furthermore, an OR was observed in 64.1% of the patients, consisting of a CR in 199 (28.9%) and a PR in 242 (35.2%). These response rates were at the lower end of the range compared to those achieved in previously published, smaller ILI series (7-9).
In the current study, the multivariate analysis identified a higher melphalan dosage as a predictor of increased grade III/IV limb toxicity. The toxic effect of melphalan varies between patients due to its lower uptake in fatty tissues compared to muscle or skin, creating a relatively higher concentration available to other tissues in patients with a high body fat percentage, resulting in a higher risk of increased limb toxicity (16). The higher body fat percentage in women and the larger limb volumes in younger patients likely accounted for the higher limb toxicity grades found in these patients, confirming results from previous studies (17,18). Similarly, the lower limb toxicity after upper limb ILI may have been the result of the higher muscle to fat ratio compared to lower limbs, however, this could also reflect the greater difficulty of delivering a therapeutic concentration of cytotoxic agents to the upper limb, resulting in reduced limb toxicities and lower OR rates (11,19,,20). Shorter ischemia times were associated with reduced limb toxicity in the univariate analysis, but not with lower response rates. This supports the hypothesis that less tissue hypoxia results in reduced tissue damage and that most melphalan uptake into the muscles takes place in the early part of the ILI procedure (17,21). On the other hand, melphalan does require an acidotic and hypoxic environment to reach its full potential (22). Earlier studies have also explored shorter procedure ILI times and, in contrast to the current study, found similar toxicity rates but lower response rates compared to longer procedure times (23). Therefore, a 30-minute procedure time is considered an effective compromise between maximising melphalan efficacy while keeping limb ischemia low, although further research will be required to explore the ideal procedure time. The results of the current study also confirmed the commonly accepted maximum effective dose of 10mg/L, above which there is no additional benefit in response to be expected, while toxicity only increases further.
In comparison to previous studies, we included patients with more advanced stages of disease (stage IIIB/C only) and with higher BOD, both reducing the likelihood of a favourable response, as these factors were identified as independent predictive factors in the current study, confirming the importance of tumour biology for response to ILI (15).
Upper limb ILI was also associated with lower OR rates in the univariate analysis, similar to the outcomes reported in a recent article specifically focusing on the topic (20). This likely reflects the technical difficulty of upper limb ILI, as upper limb procedures are considered more complex due to smaller caliber vessels, limiting venous flow and cytotoxic drug delivery (20,26). Surprisingly, univariate analysis showed that a low actinomycin-D dose was associated with a higher OR rate. This finding has not previously been reported, however, it could indicate the recently discovered synergy between low dose actinomycin-D and the tumour suppressor gene p53 (27,28). Finally, in the univariate analysis, Australian centres showed improved OR rates compared to the US centres. However, it is unclear if this is a true difference, or that it can be attributed to the different response criteria (WHO vs. RECIST) and different timing of determining tumour response used in both countries (13,14).
Some limitations of the current study have to be addressed. Firstly, we report a retrospective cohort with protocol variations between centres, most notably the differences in the response reporting criteria used between the US centres (which used the RECIST criteria) and Australian centres (which used the WHO criteria) (13,14). To minimize the potential effect this has on outcomes, only experienced high-volume centres were eligible to participate in the study and we included patients with similar stages of disease (stage IIIB/C disease) in order to exclude potential outliers. Furthermore, due to the retrospective nature of the study, we were unable to gather sufficient data on other treatments that the patients may have received before or after ILI. Lastly, as mentioned earlier, drug dosages varied between countries, as the US centres corrected melphalan dosage for ideal body weight, whereas Australian centres did not (11).
In the past decade, immunotherapy for stage IV melanoma has become an important, often effective, treatment option. However, its role for loco-regionally metastatic melanoma has not been explored specifically, with patents suffering from in-transit metastases making up only a small percentage in most studies (2-4,29). Also, since systemic side-effects of immunotherapy can be severe and even fatal, ILI for melanoma in-transit metastases should continue to be considered, providing a safe and effective treatment option without any systemic side-effects and achieving high response rates. In the future, an interesting approach to potentially optimize outcomes and minimize systemic side-effects could be to combine immunotherapy with ILI. In a study using this approach, involving 26 patients, an OR rate of 84% was achieved with Wieberdink grade III/IV limb toxicity in 27% (29).
Although these results are promising, they came at the expense of 50% of patients suffering grade 3 or grade 4 systemic adverse events. Other adjuvant systemic therapies have also been trialed in an effort to augment the effectiveness of ILI, however, in a trial in which tumour vasculature was targeted with the N-cadherin inhibitor ADH-1 following ILI, no impact on response rate was noted (30).
In conclusion, this international multi-centre study confirms that ILI is a safe and effective procedure to treat melanoma in-transit metastases. A higher melphalan dosage is predictive of greater limb toxicity, while stage of disease and BOD are predictors for response. Understanding these predictive factors will allow for better patient selection and optimization of intra-operative parameters to increase response rates, while keeping toxicity low.
References
1. Balch CM, Gershenwald JE, Soong SJ, et al. Final version of 2009 AJCC melanoma staging and classification. J Clin Oncol 2009;27:6199-206.
2. Grunhagen DJ, Kroon HM, Verhoef C. Perfusion and infusion for melanoma in-transit metastases in the era of effective systemic therapy. Am Soc Clin Oncol Educ Book 2015:e528-34.
3. Testori A, Ribero S, Bataille V. Diagnosis and treatment of in-transit melanoma metastases. Eur J Surg Oncol 2017;43:544-60.
4. Eggermont A, Blank CU, Mandala M, et al. Adjuvant pembrolizumab versus placebo in resected stage III melanoma. N Engl J Med 2018;378:1789-801.
5. Miura JT, Kroon HM, Beasley GM, et al. Long-term oncologic outcomes after isolated limb infusion for locoregionally metastatic melanoma: An international multicenter analysis. Ann Surg Oncol 2019;26:2486-94.
6. Sanki A, Kroon HM, Kam PC, Thompson JF. Isolated limb Dactinomycin perfusion and isolated limb infusion for malignant lesions of the extremities. Curr Probl Surg 2011;48:371-430.
7. Kroon HM, Huismans AM, Kam PC, Thompson JF. Isolated limb infusion with melphalan and actinomycin-D for melanoma: A systematic review. J Surg Oncol 2014;109:348-51.
8. Kroon HM, Coventry BJ, Giles MH, et al. Safety and efficacy of isolated limb infusion chemotherapy for advanced locoregional melanoma in elderly patients: An Australian multicenter study. Ann Surg Oncol 2017;24:3245-51.
9. Beasley GM, Caudle A, Petersen RP, et al. A multi-institutional experience of isolated limb infusion: defining response and toxicity in the US. J Am Coll Surg 2009;208:706-15.
10. Kroon HM, Huismans A, Waugh RC, et al. Isolated limb infusion: technical aspects. J Surg Oncol 2014;109:352-6.
11. Beasley GM, Petersen RP, Yoo J, et al. Isolated limb infusion for in-transit malignant melanoma of the extremity: a well-tolerated but less effective alternative to hyperthermic isolated limb perfusion. Ann Surg Oncol 2008;15:2195-205.
12. Wieberdink J, Benckhuysen C, Braat RP, et al. Dosimetry in isolation perfusion of the limbs by assessment of perfused tissue volume and grading of toxic tissue reactions. Eur J Cancer 1982;18:905-10.
13. Hunter RD: World Health Organization (WHO) Handbook for reporting results of cancer treatment. Geneva, Switzerland, 1979.
14. Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009;45:228-47.
15. Muilenburg DJ, Beasley GM, Thompson ZJ, et al. Burden of disease predicts response to isolated limb infusion with melphalan and actinomycin D in melanoma. Ann Surg Oncol 2015;22:482-8.
16. Cho YK, Sborov DW, Lamprecht M, et al. Associations of high-dose melphalan pharmacokinetics and outcomes in the setting of a randomized cryotherapy trial. Clin Pharmacol Ther 2017;102:511-9.
17. Kroon HM, Moncrieff M, Kam PC, et al. Factors predictive of acute regional toxicity after isolated limb infusion with melphalan and actinomycin D in melanoma patients. Ann Surgical Oncol 2009;16:1184-92.
18. Freedman DS, Jacobsen SJ, Barboriak JJ, et al. Body fat distribution and male/female differences in lipids and lipoproteins. Circulation 1990;81:1498-506.
19. McMahon N, Cheng TY, Beasley GM, et al. Optimizing melphalan pharmacokinetics in regional melanoma therapy: Does correcting for ideal body weight alter regional response or toxicity? Ann Surg Oncol 2009;16:953-61.
20. Kroon HM, Coventry BJ, Henderson MA, et al. Evaluation of the efficacy and toxicity of upper extremity isolated limb infusion chemotherapy for melanoma: An Australian multicenter study. Eur J Surg Oncol 2019;45:832-7.
21. Santillan AA, Delman KA, Beasley GM, et al. Predictive factors of regional toxicity and serum creatine phosphokinase levels after isolated limb infusion for melanoma: a multiinstitutional analysis. Ann Surg Oncol 2009;16:2570-8.
22. Beasley GM, Miura J, Zager JS, et al. Isolated limb infusion for melanoma. In: Balch CM, Thompson JF et al, Cutaneous Melanoma 6th edition, Springer nature, Switzerland, 2020 pp 827-50.
23. Huismans AM, Kroon HM, Kam PCA, Thompson JF. Does increased experience with isolated limb infusion for advanced melanoma influence outcome? A comparison of two treatment periods at a single institution. Ann Surg Oncol 2011;18:1877-83.
24. Cheng TY, Grubbs E, Abdul-Wahab O, et al. Marked variability of melphalan plasma drug levels during regional hyperthermic isolated limb perfusion. Am J Surg 2003;186:460-7.
25. Roberts MS, Wu ZY, Siebert GA, et al. Pharmacokinetics and pharmacodynamics of melphalan in isolated limb infusion for recurrent localized limb malignancy. Melanoma Res 2001;11:423-31.
26. Beasley GM, Sharma K, Wong J, et al. A multi-institution experience comparing the clinical and physiologic differences between upper extremity and lower extremity melphalan-based isolated limb infusion. Cancer 2012;118:6136-43.
27. Choong ML, Yang H, Lee MA, et al. Specific activation of the p53 pathway by low dose actinomycin D: a new route to p53 based cyclotherapy. Cell Cycle 2009;8:2810-8.
28. Tzaridis T, Milde T, Pajtler KW, et al. Low-dose actinomycin-D treatment re-establishes the tumour suppressive function of P53 in RELA-positive ependymoma. Oncotarget 2016;7:61860-73.
29. Pasquali S, Hadjinicolaou AV, Chiarion V, et al. Systemic treatments for metastatic cutaneous melanoma. Cochrane Database of Syst Rev 2018:CD011123.
30. Beasley GM, Riboh JC, Augustine CK et al. Prospective multicenter phase II trial of systemic ADH-1 in combination with melphalan via isolated limb infusion in patients with advanced extremity melanoma. J Clin Oncol 2011;29:1210-5.