Phase 2 Study of Intralesional PV-10 in Refractory Metastatic Melanoma

Eighty patients were enrolled onto this international, multicenter, open-label, single-agent study. The governing institutional ethics committee for each study center approved the study protocol, and all patients provided written informed consent. Eligible patients had biopsy-proven confirmation of melanoma and at least one cutaneous or subcutaneous lesion ≥0.2 cm in diameter that could accurately be measured by ruler/caliper or ultrasound. They received a single IL injection of PV-10 to uniformly infiltrate each of up to 20 study lesions on day 0 (i.e.,≤10 target and ≤10 nontarget dermal lesions) using 0.5 mL PV-10 per cm³ of lesion volume. Treatment could be repeated at weeks 8, 12, and 16 for new nontarget lesions or existing target or nontarget lesions not exhibiting complete response. PV-10 was not injected into nodal or visceral lesions. One or two additional measureable, biopsy-confirmed dermal lesions could be designated for assessment of bystander response and were followed but not injected. Body mapping and digital photography with lesion identification markers and reference scale were used to accurately track all study lesions. Patients were followed for 52 weeks after initial PV-10 treatment (i.e., for at least 36 weeks after their last possible PV-10 treatment).

Study evaluations, including lesion photography, were performed at screening, on the day of PV-10 administration, at 1 day and 7 days after injection, and every 4 weeks thereafter during the treatment portion of the study (i.e., through week 16), and during long-term follow-up at weeks 24, 36, and 52. Radiologic assessments of visceral disease status were performed every 12 weeks throughout the study, and patients were transitioned into survival follow-up if at any time the investigator identified clinical or radiologic evidence of distant progression. No other melanoma therapy was permitted during the study interval. Treatment for concurrent or intercurrent illness, and wound care or management of pain, were allowed at each investigator’s discretion. Adverse events (AEs) were monitored over the study interval.

The study utilized a modified Fleming two-stage design.14 Interim assessment upon accrual of the first 40 patients required an objective response in at least eight patients to substantiate a true response rate between 10 and 30 %. Full accrual allowed testing of a projected 30 % primary efficacy rate (95 % confidence interval 20–40). Post hoc exploratory analyses were undertaken to understand results in certain subgroups.

The primary end point of the study was BORR of target lesions, with secondary end points of progression-free survival (PFS) and duration of response, overall survival (OS), by-lesion BORR, and bystander lesion BORR, along with assessment of AEs, quality-of-life (assessed using the EORTC QLQ-C30 instrument), lesion pain (assessed via a 100 mm visual analog scale), and pharmacokinetics.15‐17

Response Evaluation Criteria in Solid Tumors (RECIST) was modified (mRECIST) to allow: (1) designation of cutaneous or subcutaneous lesions ≥0.2 cm in diameter as target lesions; (2) up to 10 cutaneous and subcutaneous target lesions to be followed; and (3) assessment of disease progression against baseline sum of longest diameters of target lesions, thereby allowing clinically insignificant progression (including locoregional appearance of new cutaneous or subcutaneous nontarget lesions) to be treated to week 16.18 Standard thresholds for percentage change in sum of longest diameters were used to define complete response (CR), partial response (PR), stable disease (SD), and progressive disease (PD). All lesions specified at baseline were followed over the course of the study with last observations carried forward for any lesions not measurable at a visit; response assessment was censored for missed visits. When eschar was reported, a standard query was used to ascertain lesion status. An example of measurable eschar that eclipsed baseline measurement is illustrated in Fig. 1a. To mitigate such artifacts and allow detection of regrowth after initial ablation, the first formal response assessment occurred at week 8 and was repeated at each visit thereafter. Patients failing to reach assessment at week 8 were deemed not evaluable and classified as PD.

Eighty patients (intent-to-treat population, ITT; median age 70 years, range 33–97 years; Table 1) were enrolled over 19 months at 7 study centers: 3 in Australia and 4 in the United States. All patients had recurrent, locally advanced melanoma after a median of 6 previous interventions (range 1–19), and most had received multiple classes of treatment. The median time from initial melanoma diagnosis to initial PV-10 treatment was 42.3 months (range 1.7–752). Study lesions comprised a substantial total tumor burden with a median sum of longest diameters of 6.3 cm (range 0.9–21.0 cm), with a median of 7.5 study lesions per patient (range 1–22). The most prevalent comorbidity was hypertension (51 % of patients), followed by hypercholesterolemia (23 %) and depression (19 %).

Patients received a mean of 1.8 PV-10 treatment cycles (median 2); 35 patients received a single cycle, and three patients received the maximum four cycles allowed by protocol.

Half of the ITT patients achieved an objective response in their target lesions (51 % BORR, 26 % CR + 25 % PR) (Table 2), with 8 % of patients having no evidence of disease after 52 weeks. Locoregional disease control (SD or better) was achieved in 69 % of patients, while 16 % were not evaluable as a result of progression before week 8. (This occurred predominantly in patients with extensive disease burden not injected with PV-10.) Onset of response occurred at a median of 1.9 months, corresponding with the first assessment of response. The median duration of response was 4.0 months (by RECIST) but was not met in the study interval when assessed by mRECIST. Time-to-event data for objective response are represented in Fig. 2, and PFS data for all patients are summarized in Table 2.

Response rates by lesion among 491 target lesions in the ITT population were similar to those by patient, with 53 % of lesions achieving CR, 5 % PR, and 12 % SD.

Among the 42 patients with designated bystander lesions, 26 % experienced CR of their uninjected bystander lesions, 7 % PR, and 17 % SD. Response in bystander lesions strongly correlated with that of patients’ target lesions: CR or PR in target disease was associated with 56 % CR and 6 % PR in bystander lesions; in contrast, patients who did not experience an objective response of their target lesions had 6 % CR and 12 % PR of their bystander lesions (P = 0.023, BORR by χ ² test).

Median OS was not reached for the ITT population. Mean OS for stage III patients was >12 months (89 % 1-year survival, median not reached), while for stage IV patients, median survival was 6.5 months (39 % 1-year survival).

Because the protocol limited PV-10 treatment to the first 16 weeks, 11 patients subsequently withdrew to receive further PV-10 treatment under alternative protocols.

All patients experienced one or more AE during the study (Table 3); most were grade 1 or grade 2, while 15 % of patients had at least one grade 3 AE deemed at least possibly related to study treatment. The most common AEs occurred at the injection site, led by transient pain (reported by 80 % of patients), edema (41 %), and vesicles (39 %). Six patients experienced mild (4 %) or moderate (4 %) injection site photosensitivity, and one (1 %) experienced a severe generalized photosensitivity reaction. All of these resolved without sequelae. No life-threatening or fatal AEs at least possibly related to the study treatment were reported.

Quality-of-life assessment throughout the study interval showed no substantial change from baseline on any quality-of-life dimension after PV-10 treatment. Pain scores indicated transient pain in treated lesions during the first week after treatment that generally resolved to baseline within 4 weeks. Approximately 60 % of patients received local anesthesia at the time of PV-10 injection.

Pharmacokinetic data were consistent with the literature on RB and illustrate PV-10 clearance via an apparent biexponential process, with a rapid initial distribution/absorption phase occurring during the first 24 h (k_D/A = 0.0020 min⁻¹, t_1/2,D/A = 5.9 h), followed by slower terminal elimination (k_E = 0.00012 min⁻¹, t_1/2,E = 100 h).19 Relatively low uptake during the initial phase (geometric mean  %AUC∞ = 17.3 %) and intercepts for the initial and terminal phases (427 and 73 ng/mL, respectively) are consistent with prolonged retention of drug in injected lesions, with potentially significant systemic exposure only during the initial phase.

Although only cutaneous and subcutaneous lesions were injected, the study enrolled patients with stable visceral disease (including brain, lung, and liver metastases) who also had injectable dermal disease. To test for a relationship of response to untreated disease burden, patients were classified according to their reported tumor burden at baseline: all known disease treated with PV-10 (28 patients); all known disease treated with PV-10 (with the exception of 1–2 uninjected bystander lesions) (26 patients); 10 or fewer untreated and unmeasured dermal lesions (7 patients); and dermal disease burden classified as too numerous to count or as stage IV disease (19 patients). For the 54 patients in the first two subgroups, their study lesions, including bystander lesions, represented all documented disease at baseline. These patients, who received PV-10 to all or substantially all of their disease burden, achieved markedly higher response rates compared to patients with substantial untreated disease burden (Table 2).

Another exploratory analysis was undertaken to understand the relevance of locoregional vesicles (blistering). Transient, fluid-filled bullae were observed in 40 % of patients (23 % grade 1, 16 % grade 2, and 1 % grade 3, including one occurrence reported as possible grade 1 photosensitivity), both perilesional and locoregional and treatment emergent within 1–7 days of PV-10 administration. Onset was independent of dose with no apparent relationship to prior PV-10 administration (Fig. 1). These generally resolved within 4 weeks, with or without medication intervention and without long-term sequelae; blistering was associated with a marked increase in response: 44 % of patients with blisters experienced CR, versus 15 % without blisters (P = 0.008).

Exploratory analyses were undertaken to assess relevance of demographics, treatment history, and investigator. Equivalent target lesion responses occurred in patients above and below the median age (54 vs. 49 % BORR, P = 0.8) and for men and women (58 vs. 41 %, P = 0.2). Dichotomization according to treatment history (systemic or regional chemotherapy or immunotherapy vs. naive patients) showed no significant difference in target lesion response (43 vs. 58 %, P = 0.3). Similar response rates were observed across study centers, with all centers reporting at least one patient experiencing PR or better (Fig. 2), and with five of the seven centers reporting patients with no evidence of disease after 12 months. Patterns of AEs were also similar across centers.

Although the study was not designed to quantitatively follow lesions in visceral organs, comprehensive radiologic imaging provided some insight into whether PV-10 could have an impact on visceral disease. While a substantial fraction patients with stage IV disease were classified as not evaluable due to early progression (Table 2), four patients experienced SD or PR of their visceral disease (including patients with multiple pleural and hepatic metastases); three of these patients also exhibited SD or better outcome in their target lesions, similar to the correlation of bystander response to that of target lesions. Seven stage IV patients survived through the end of the 52-week follow-up interval, including 4 of 10 patients with M1c disease.