Patients were eligible for this trial if they were 18 years of age or older with measurable metastatic melanoma and a clinical performance status of 0 or 1 on the Eastern Cooperative Oncology Group (ECOG) scale and life expectancy of greater than 3 months. Patients were excluded from participation if blood counts were outside the normal range, if systemic steroid therapy was required, or if patients had other active systemic infections, coagulation disorders or other active major medical or cardiovascular diseases or immunodeficiencies, or were positive for HIV antibody or hepatitis B or C. All patients signed an informed consent approved by the institutional review board of the National Cancer Institute.

We previously reported the results of 35 patients who received TIL transfer after a NMA lymphodepleting regimen consisting of cyclophosphamide 60 mg/kg/d for 2 days and fludarabine 25 mg/m²/d for the next 5 days (Fig 1A). Two additional trials were instituted with the end points of evaluating improved response rates and survival. In the next pilot trial, 25 patients received this chemotherapy over 5 rather than 7 days followed by a single fraction of 2 Gy of TBI. In a third trial, 25 patients received the same 5 days of chemotherapy plus 12 Gy of TBI administered as 2 Gy twice a day for 3 days (Fig 1A). Some patients who met immunologic and clinical criteria were not eligible to receive TBI because of prior radiation therapy or an inability to collect adequate CD34⁺ stem cells. Eight additional patients were subsequently treated with the original NMA regimen without TBI, and the results in this entire cohort of 43 NMA-only patients are updated here with a median follow-up of 45 months.

For the 2-Gy TBI regimen, patients underwent clinical simulation for TBI that included measurement of maximum separation, typically at mid-hip. TBI was delivered with the patient positioned supine on a gurney with arms placed by the sides for lung compensation. No additional tissue compensators were used. Treatment was delivered at 2-Gy fractions with 15-MV photons delivered at a distance of 6 meters to the patient's midline at a dose rate of 0.11 Gy/min. For the 12-Gy TBI regimen, a computed tomography simulation was performed to allow planning for head, neck, and lung compensators. TBI was delivered with opposed lateral fields as described for the 2-Gy regimen with the addition of compensation. Mediastinal boosting of tissues underdosed as a result of lung compensation was accomplished with anterior-posterior and posterior-anterior 6-MV beams to obtain a final dose of 12 Gy to the total body and mediastinum while limiting the median lung dose to 6 Gy. All fields were treated twice daily with at least a 6-hour interfraction interval over 3 consecutive days. All patients received a single dose of ondansetron 0.15 mg/kg intravenously before delivery of TBI as nausea prophylaxis.

On the day after the final dose of chemotherapy or TBI, patients received a bolus intravenous infusion of TIL over 0.5 to 1 hour, and started high-dose IL-2 within 24 hours (720,000 U/kg intravenously every 8 hours to tolerance up to 15 doses). One or 2 days after TIL infusion, a minimum of 2 × 10⁶/kg of autologous purified (Miltenyi Biotech, Bergisch Gladbach, Germany) cryopreserved CD34⁺ hematopoietic stem cells from a granulocyte colony-stimulating factor–mobilized pheresis were infused intravenously in only patients who received TBI.

After TIL infusion, all patients received trimethoprim, sulfamethoxazole, and fungal prophylaxis; herpes virus–seropositive patients received valacyclovir. Patients were monitored daily after TIL infusion and received platelets or packed RBCs as needed. Patient response was assessed using standard radiographic studies and physical examination at 4 weeks after cell administration and at regular intervals thereafter, and were categorized into complete, partial, or nonresponders based on Response Evaluation Criteria in Solid Tumors (RECIST) guidelines.

TIL cultures for patient treatment were generated as previously described.^3,13 The phenotype of TIL was determined by fluorescence-activated cell sorting (FACS) analysis and the antigen specificity was assessed by cytokine-release assays using standard methods.^3,13 The mean length of telomere repeats at chromosome ends in individual T-cell cultures was measured by quantitative flow-fluorescent in situ hybridization as previously described.¹⁴

Serum was collected from patients and frozen before initiation of therapy and after lymphodepletion but before cell infusion. Sera were thawed and assayed at the same time in a blinded fashion using commercially available enzyme-linked immunosorbent assay kits.

Standard statistical methods are cited in the text, and all P values are two-tailed assuming unequal variance, with P < .05 considered significant.

The characteristics of the patients enrolled onto these three trials are shown in Table 1. Patients in the three clinical trials received a similar number and quality of TIL that had been selected for antitumor reactivity and then rapidly expanded to large cell numbers and infused immediately after collection.

The objective antitumor responses and durations of response are summarized in Table 2. The response rates in the NMA trial (n = 43), the NMA plus 2-Gy trial (25 patients), and the NMA plus 12-Gy trial (n = 25) were 48.8%, 52.0%, and 72.0% respectively. Tumor regression was seen in metastases at virtually all visceral and soft tissue sites including brain. Thus, the high response rates seen in the additional 58 patients reported here who received ACT with the NMA with or without additional TBI conditioning extend and confirm the high objective response rate reported by us earlier in 35 patients who received TIL transfer after the NMA conditioning regimen.³ Among these 93 patients, 34 patients (37%) had received prior chemotherapy (the response rate in this group was 62%) and 77 patients (83%) had received prior IL-2 therapy (the response rate in this group was 56%). Among the 10 patients who achieved a complete response, there have been no relapses with a median follow-up of 31 months. According to the American Joint Committee on Cancer (AJCC) staging system, of the 10 patients who experienced a complete response, two patients were stage M1a (metastases to cutaneous, subcutaneous, and nodal sites only), four patients were stage M1b (metastases to the lung), and four patients were stage M1c (metastases to other visceral and distant sites).

The median follow-up as of January 1, 2008, for the three trials was 45, 27, and 10 months, respectively. The 2-year survival rates were 30% and 42%, respectively, in the trials involving no TBI or 2-Gy TBI (Fig 1B), and follow-up was too short in the 12-Gy trial. In these sequential studies, there was no significant difference in survival among the three treatment groups (P = .13), although more prolonged follow-up is necessary.

The characteristics of the treatments administered to patients are shown in Table 3. All infused TIL demonstrated specific recognition of autologous melanoma or human leukocyte antigen–A2 matched melanoma lines by cytokine-release assays (Appendix Tables A1-A3, online only). Multiple features of the infused TIL cultures were evaluated for correlation with patient response, but the only treatment characteristic that was significantly correlated with clinical outcome was the number of IL-2 doses tolerated by patients. The responding patients in the 2-Gy trial tolerated significantly less IL-2 (6.9 doses) than nonresponding patients (9.3 doses; P =.01). The 52 responding patients from all three clinical trials tolerated significantly less IL-2 than the 41 nonresponding patients (P < .001). These results could suggest that TIL that are effectively engaged in antitumor responses are releasing secondary cytokines into the host's circulation that limit IL-2 tolerance compared with TIL in noneffective interactions.

Patients exhibited the expected hematologic toxicities associated with the cyclophosphamide, fludarabine, and TBI preparative regimens. Patients recovered marrow function rapidly after cell infusion with absolute neutrophil counts greater than 500/mm³ by day 12 and sustained platelet counts greater than 20,000/mm³ by day 14 (except for four patients on the 12-Gy TBI protocol with platelet recovery on days 16, 17, 20, and 22). Patients conditioned with 12-Gy TBI demonstrated on average a 1- to 2-day delay in marrow recovery compared with other protocol patients (Appendix Fig A1, online only) and required significantly more blood-product support than patients treated with 2-Gy TBI (Table 4; P for platelets = .002; P for RBC = .05). Patients experienced the usual reversible and transient toxicities associated with IL-2 that have been extensively described elsewhere.¹⁵ The significant grade 3 and 4 treatment-related toxicities that were not associated with IL-2 administration in the two TBI protocols are shown in Table 4. Not surprisingly, both the spectrum and the intensity of toxicities were less severe in these patients who received autologous cells than those of toxicities reported in patients undergoing myeloablative conditioning regimens with allogeneic stem-cell transplantation and chronic immunosuppression. For instance, among the 50 patients who received TBI conditioning, only one patient, who received 2-Gy TBI, experienced significant pulmonary toxicity, consisting of pulmonary veno-occlusive disease refractory to steroids. This patient subsequently responded to sildenafil therapy, improved, and returned to work as a nurse, although she sometimes requires oxygen by nasal cannula when exerting. One patient who received 2-Gy TBI and four patients who received 12-Gy TBI required intubation for somnolence. Four patients who received 12-Gy TBI developed renal dysfunction attributed to thrombotic microangiopathy with elevated serum creatinine levels, though those have stabilized. There was one treatment-related death in the 2-Gy TBI protocol resulting from sepsis in a neutropenic patient who had a pre-existing but unrecognized diverticular abscess at protocol enrolment. One patient in the 12-Gy TBI group developed grade 3 posterior uveitis that was refractory to topical steroids. MART-1 tetramer-positive lymphocytes were recovered from the patient's ocular vitreous fluid during an acute uveitis phase accompanied by visual changes. The patient's uveitis symptoms resolved after intraocular steroid injections and treatment with diazoxide.

As in our prior report of patients receiving ACT with the lymphodepleting chemotherapy regimen alone, there was no significant correlation between a large number of phenotypic markers in the infused lymphocytes and clinical response. There was, however, a strong correlation between the likelihood of response across all preparative protocols and the mean telomere length of the infused cells (Fig 2A). This observation suggest that the proliferative potential of infused TIL is an important attribute of their ability to mediate tumor destruction in the patient.

The role of serum cytokines in the antitumor efficacy of lymphocytes administered after lymphodepletion was investigated. Serum was sampled before initiation of the lymphodepleting preparative chemotherapy and immediately before cell administration (after lymphodepleting therapy), and circulating levels of the T cell homeostatic regulatory cytokines, IL-7, and IL-15 were quantified by enzyme-linked immunosorbent assay (Fig 2B). The starting levels of homeostatic cytokines were similar for all three populations studied. Lymphodepletion caused an elevation of the levels of IL-7 and IL-15 for all protocols. The ratio of the final concentration to the initial concentration of serum cytokines increased significantly more when patients received 12-Gy TBI than when they received NMA chemotherapy alone (P for IL-7 = .02; P for IL-15 = .005).