Phase IB/II Study of Second-Line Therapy with Panitumumab, Irinotecan, and Everolimus (PIE) in KRAS Wild-Type Metastatic Colorectal Cancer

The management of metastatic colorectal cancer (mCRC) has evolved over the past decade. There have been significant improvements seen in survival when patients are treated with all chemotherapy agents, and further improvements are noticed with the addition of biological agents (1–3). More recently, there has been ongoing refinement of colorectal cancer into biological subgroups with classification into RAS wild-type (WT), RAS mutant (MT), and BRAF MT groups (4).

In patients with RAS WT mCRC, the addition of an EGFR inhibitor to irinotecan schedules improves progression-free survival (PFS) in second-line therapy (5–7). Activation of the EGFR results in cell proliferation and survival via the RAS/RAF/MAPK and PI3K/PTEN/AKT signaling pathways. Patients with RAS MT have constitutive activation of the RAS/RAF/MAPK pathway and are resistant to therapy with an EGFR inhibitor (4). Additional mutations within these pathways may also result in resistance (8).

mTOR is a key downstream regulatory protein, activated via the PI3K–AKT pathway, and regulates cell growth, proliferation, and survival. In preclinical studies, the addition of the mTOR inhibitor everolimus to an EGFR inhibitor restored cell-survival inhibition in resistant cancer cell lines and showed cooperative antitumor activity in xenograft models, suggesting that inhibition of mTOR in addition to blockade of the EGFR may overcome upstream mechanisms of resistance (9). The addition of an mTOR inhibitor to anti-EGFR therapy in KRAS WT colorectal cancer could therefore improve outcome in second-line therapy.

Two prior studies of the combination of cetuximab, irinotecan, and everolimus were conducted before testing for KRAS MT had become routine and both studies were stopped early (10, 11). The largest of these did not show any activity in patients with KRAS MT disease but showed a response rate of 20% and final PFS of 5.6 months in KRAS WT patients receiving second-line therapy (11). This phase Ib/II study assessed the combination of panitumumab, irinotecan, and everolimus (PIE) in purely KRAS WT exon 2 mCRC in second-line therapy.

This was a prospective, single-arm study conducted in five Australian sites, of the combination of PIE in patients with mCRC with KRAS WT following failure of first-line fluoropyrimidine (5FU)-based chemotherapy. The study was conducted in two parts, the initial phase Ib dose-finding component for which the primary endpoint was MTD, followed by the phase II expansion study for which the primary endpoint was response rate.

This study was performed after approval by the local human ethics committees at each site, and informed consent was obtained from each participant for participation and translational research. The study was conducted in accordance with the Declaration of Helsinki.

Eligible patients had KRAS exon 2 WT (local laboratory testing) mCRC and must have failed first-line fluoropyrimidine-based chemotherapy (radiologic progression, development of metastatic disease within 6 months of completing adjuvant therapy, or toxicity to first-line therapy limiting further treatment). Patients who had received first-line anti-EGFR therapy were excluded. Patients were required to be over 18 years old, able to give informed consent, and have Eastern Cooperative Oncology Group (ECOG) of 0 to 1 for the phase Ib group, with ECOG of 0 to 2 allowed for the phase II component. Patients were required to have adequate bone marrow, renal, and hepatic function, fasting cholesterol ≤ 7.75 mmol/L, and fasting triglycerides ≤ 2.5 x upper limit of normal (ULN). Patients with poorly controlled diabetes mellitus, untreated or symptomatic CNS metastases, or other severe and/or uncontrolled medical conditions were excluded. Patients who had received prior pelvic radiotherapy were excluded from the phase Ib study.

On day 1, patients received panitumumab as a 60-minute i.v. infusion, followed by irinotecan given as a 90-minute i.v. infusion which was repeated every 14 days. Everolimus was administered orally as a daily dose. The dose levels used are shown in Table 1. Dose level 1 was the starting dose and was the dose used in the expansion phase. Treatment continued until disease progression or unacceptable toxicity. If one agent was discontinued due to toxicity, the remaining components of therapy could be continued.

Adverse events were collected throughout treatment and until 30 days following last treatment and were graded using the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE v3.0). Tumor response was assessed by investigators every 6 weeks using the RECIST v1.0 as assessed by CT scan of the chest, abdomen, and pelvis. Confirmatory CT scan was not mandatory in the protocol. The end of study visit occurred 30 days following completion of treatment, and patients were then followed up every 3 months until death or lost to follow-up.

Primary endpoint of the phase Ib study was the MTD of PIE when given in combination where the MTD was the dose with a dose-limiting toxicity (DLT) occurring in ≤1 of 6 patients. Dose finding used a standard 3+3 design, with patients enrolled in cohorts of a minimum of three evaluable patients [evaluable defined as either completing two cycles (28 days) of treatment or experiencing a DLT]. During the dose-escalation phase, if there was no DLT in the first three patients, the next cohort proceeded at the next dose level. Where a DLT occurred in one patient, the cohort was expanded to include six patients. If a DLT occurred in two or more patients from the expanded cohort, the MTD had been exceeded, and the prior dose level was to be declared the MTD. The predefined DLTs included any of the following occurring within 28 days of beginning therapy: febrile neutropenia, grade 4 thrombocytopenia, grade 3/4 neutropenia taking more than 14 days to resolve, any grade 3 nonhematologic event taking more than 7 days to resolve, any grade 4 nonhematologic event, and any treatment delays or omissions of 14 days or longer.

The primary endpoint of the phase II/expansion study was response rate as determined by RECIST v1.0, and secondary endpoints were safety and toxicity, PFS, and overall survival (OS). All analyses from the phase II study were intention to treat. PFS and OS were analyzed using the Kaplan–Meier method.

A response rate of 35% would be sufficient to investigate this combination further and therefore based on Simon two-stage Minimax design using a 20% RR with standard care, a sample size of 53 patients would have 80% power with 95% confidence to detect a RR consistent with 35% [95% confidence interval (CI), 22%–48%], ranges from 22%–48% (http://faculty.vassar.edu/lowry/prop1.html).

Formalin-fixed, paraffin-embedded (FFPE) samples of tumor tissue from archival specimens were retrieved from storage at hospital pathology departments. Genomic DNA was extracted from 10-μm-thick tissue sections (1–2) using the QIAamp DNA FFPE tissue Kit (Qiagen). Manual microdissection was performed on samples with less than 80% malignant cells when visualized by microscopy. Whole-exome sequencing was performed at the ACRF Cancer Genomics Centre (Adelaide). Briefly, whole-exome capture libraries were constructed from 200 to 500 ng of DNA. Six sample libraries were pooled and subjected to hybrid capture using Roche SeqCap EZ Human Exome Library v3.0 and xGEN Universal Blocking Oligos (Integrated DNA Technologies). Sequencing of each capture pool was performed using 2 × 100 paired-end reads on 1 lane of an Illumina HiSeq 2500. Reads were mapped to the hg19 reference genome, and variants called using the GATK tool. Filtering for variants in genes associated with progressive disease was performed. The Fisher exact test was used to determine variants significantly associated with response. OncoFOCUS panel (Agena Bioscience) was used for extended RAS in cases with insufficient DNA for whole-exome sequencing. Kaplan–Meier survival curves and Cox regression analyses were performed using Graph Pad Prism v6 and SPSS v18 9 (PASW Statistics for Windows, Version 18.0. Chicago: SPSS Inc.).

Forty-nine patients were enrolled between June 30, 2010, and June 17, 2015. The flow chart is represented in Supplementary Fig. S1. There were 11 evaluable patients in the phase Ib component of the study, and an additional 34 patients enrolled in the expansion phase, with a total of 40 patients treated at the MTD. One patient in the expansion phase deteriorated prior to receiving any therapy. Results of the phase II study were analyzed as intention to treat. The study was closed early following enrolment of 40 patients in the phase II study due to slow recruitment. Baseline patient demographics and disease characteristics of evaluable patients are shown in Table 2.

Three patients were treated in cohort 1 at dose level 1 with no recorded DLT. Cohort 2 then enrolled three patients at dose level 2. One of three patients experienced a DLT of grade 3 mucositis, taking more than 7 days to resolve to grade 1. Cohort 2 was subsequently expanded to include six patients. Following treatment of two of the additional three patients, a further DLT of grade 3 mucositis taking more than 7 days to resolve occurred, resulting in two of five patients experiencing a DLT at dose level 2. Cohort 2 was suspended prior to the 6th patient receiving any therapy. A further three patients were then enrolled at dose level 1 with no DLT observed in a total of six patients treated at dose level 1. The MTD was declared as dose level 1.

The adverse events experienced in the phase II study are shown in Table 3. And the mean number of cycles of all three agents received was 8.2. The median relative dose intensity received in the first 4 months of therapy was 85%, and four patients ceased everolimus within 30 days due to toxicity. A total of 63% of patients experienced any grade 3 or higher toxicity with 25% grade 3 neutropenia, 23% grade 3 diarrhea, 18% grade 3 mucositis, and 13% grade 3 rash. Grade 4 toxicities were hypomagnesemia (5%) and neutropenia (3%).

The overall response rate of the combination was 48%, with additional 43% experiencing stable disease as best response. Figure 1A is the waterfall plot demonstrating change in tumor volume at best response; as two patients had rapid clinical disease progression, one prior to beginning therapy and one shortly after commencing therapy, and did not undergo repeat tumor measurements, so could not be included in the waterfall plot. Two patients had sufficiently good response to allow subsequent resection of liver metastases. In patients with right-sided primary tumor location, the overall response rate was 25% (3/12), and the response rate was 55.6% (15/27) in patients with a left-sided primary tumor. The duration of response is shown in Fig. 2.

The median PFS was 5.6 months (95% CI, 4.1–7.1; Fig. 3A), and the median OS was 10.8 months (95% CI, 8.1–13.5; Fig. 3B). The 12-month OS was 48%, and 18-month OS was 33%.

Tissue for preplanned biomarker studies was available in 33 of the 40 patients. DNA was extracted from FFPE for whole-exome sequencing, including repeat KRAS exon 2 mutation testing. Four patients had additional RAS mutations: one had KRAS exon 2 G12V, two had KRAS exon 3 Q61H, and one had NRAS exon 2 G12 D. Of those with RAS mutations, one of four had a partial response (G12V mutation), and three of four had stable disease. Four patients had BRAF V600E mutation, and of these, two of four had stable disease, and two of four had disease progression; one of these had rapid clinical progression prior to repeat CT assessment. The response rate in the 33 patients was 16 of 33 (48%), and best tumor response is shown in the waterfall plot in Fig. 1B. The median PFS was 5.3 months (95% CI, 3.0–7.6), and the median OS was 10.6 months (95% CI, 7.9–13.4; Supplementary Fig. S2). In the all RAS/RAF WT group (n = 25), the response rate was 60% (15/25), the median PFS was 6.4 months (95% CI, 4.1–8.6), and median OS was 11.8 months (95% CI, 9.3–14.4; Fig. 3C and D).

This study shows that the combination of PIE is feasible with the most frequent toxicities being diarrhea, mucositis, rash, and fatigue. Although the rate of any grade 3 toxicity was high (63%), the individual toxicities were manageable, and the dose intensity of treatment was 85%. Importantly, this study also demonstrated a favorable response rate in second-line therapy with a partial response rate of 48% in KRAS WT patients and 60% in the smaller RAS/RAF WT group, showing that this is an active combination.

In this combination, we used standard dosing of panitumumab at 6 mg/kg every second week. For ease of administration, irinotecan was also administered every 2 weeks. Although a phase II trial of four different single-agent irinotecan schedules recommended a dose of 250 mg/m² of irinotecan every 2 weeks (12), we selected a starting dose of irinotecan 20% below this in view of combination therapy. Everolimus was administered at 5 mg alternate days; however, for future studies, this could be given at 2.5 mg daily for ease of scheduling. To our knowledge, there have been two other studies of irinotecan in combination with an anti-EGFR antibody and everolimus (10, 11). Both of these studies used cetuximab, irinotecan, and everolimus and showed comparable rates of toxicity as seen in our study. In the first phase I study (10), 19 patients received weekly everolimus in combination with weekly cetuximab and irinotecan administered every 3 weeks. Fourteen patients were treated at the recommended phase II dose, and of these, 64% experienced any grade 3 or higher adverse event, including neutropenia, stomatitis, and hypokalemia. Overall, this study reported slightly lower rates of diarrhea, rash, and mucositis than seen in our study but higher rates of nausea. In the second phase Ib/II study (11), 43 patients were treated with irinotecan given on days 1 and 8 of a 21-day cycle, cetuximab on days 1, 8, and 15, and daily everolimus. Similar to that seen in our study, the grade 3 toxicities were diarrhea (35%), mucositis (21%), and rash (19%). Overall, fatigue, nausea, diarrhea, rash, and stomatitis were the most common toxicities seen in the three trials.

With improved familiarity with the regimen and increased use of prophylactic measures, it may be possible to reduce the rate of grade 3 toxicity. For example, the use of prophylactic skin treatment using the STEPP protocol was recommended but not mandated as part of the protocol, and this has been shown to reduce grade 2 or higher skin toxicities by more than 50% (13). Recent studies have also reported a significant reduction in the rate of everolimus-related mucositis with the use of prophylactic dexamethasone mouth wash, with rates of grade 2 or higher mucositis occurring in only 2.4% of patients using dexamethasone prophylaxis as compared with 33% reported in prior studies (14). The high rate of diarrhea (23% grade 3) seen with this combination remains a concern. Although this was manageable with dose delays and reductions, patient education is also important to ensure appropriate use of antidiarrheal medication. In this study, we did not investigate lower irinotecan dosing (180 mg/m²), but this could be considered in future to improve diarrhea rates.

This study demonstrated a favorable response rate in second-line therapy with a partial response rate of 48% in KRAS WT patients and 60% in the smaller RAS/RAF WT group, showing that this is an active combination. In addition, two patients were able to undergo resection of liver metastases following treatment. This is favorable compared to the response rate seen in prior studies. Panitumumab, irinotecan, and infusional 5FU resulted in an RR of 35% in KRAS WT patients in the 181 trial (5) and a 45% overall response in all RAS WT (15). Thirty-four percent of KRAS WT patients receiving irinotecan and panitumumab in the PICCOLO trial had a partial response (7). The response rate to cetuximab and irinotecan of 16.4% seen in the EPIC trial cannot be compared with our findings as this trial was conducted prior to RAS testing being performed (6).

However, the partial response rate seen in our study is higher than that seen in the two prior studies of cetuximab, irinotecan, and everolimus. This may be because these studies were conducted prior to KRAS testing being standard. Hecht and colleagues did not perform KRAS testing and reported a response rate of 14% (10), whereas Chiorean and colleagues identified 20 KRAS WT patients with an RR of 20% in this group (11). The discrepancy between this result and our finding of 48% may reflect the small size of all three studies. Other potential confounders are the lack of independent radiology assessment in our study and lack of confirmatory assessment.

Our results of a median PFS of 5.6 months in all patients and 6.4 months in the all RAS/RAF WT group are very comparable for second-line therapy. Certainly, this is comparable with the CIE study which had a median PFS of 5.6 months in KRAS WT patients (11), and with the PFS of 5.9 months in KRAS WT patients and 6.4 months in RAS WT population in the 181 study (15). The benefit of the addition of everolimus to irinotecan and panitumumab is uncertain; however, the results seen here are at least comparable with those from the 181 study but without the need for infusional 5FU which may provide improved convenience for patients.

Although this was a small, uncontrolled phase II trial, a number of patients with poor prognostic factors were enrolled in this study which may have affected the survival results. Fifteen percent of patients had developed metastatic disease within 6 months of completing adjuvant chemotherapy, 10% of patients had bone metastases, and 10% had peritoneal disease. While 30% of patients had a right-sided primary lesion, this is comparable with other studies (16). At baseline, >50% of patients, reflecting a relatively high proportion compared with other studies, had elevated neutrophil-to-lymphocyte ratio. Lactate dehydrogenase (LDH) was also elevated in 58% of patients. Both markers have been associated with poor outcome, and this may have been reflected in the OS of this study.

Of the 40 patients in the phase II component of the study, we were able to perform extended RAS and BRAF testing on 33 patients. Four patients had BRAF V600E MT, and none of these patients had an objective response. Four additional RAS MT were identified, and one, a KRAS exon 2 G12V MT, was not detected on local laboratory testing. This may reflect increased detection of mutations with the higher sensitivity obtained with next-generation sequencing platforms such as OncoFOCUS. Of these four RAS MT patients, there was one objective response. Although these results do not support ongoing study in patients with BRAF or RAS mutations, the benefit of this combination in RAS WT patients who initially respond to anti-EGFR therapy but subsequently become resistant is unknown, and future study in this patient population would be of interest.

This study showed that the combination of PIE was feasible with majority of patients experiencing fatigue, rash, diarrhea, and mucositis; however, the toxicities were manageable. The efficacy outcomes are interesting and at least comparable with that seen with irinotecan and anti-EGFR combinations. The additional impact of everolimus is hard to define in this small patient population, and larger randomized studies would be required to address this further. Future studies could also assess this combination in patients who are refractory to anti-EGFR therapy.

N. Tebbutt is a consultant/advisory board member for Amgen. C. Karapetis is a consultant/advisory board member for Amgen. N. Singhal reports receiving speakers bureau honoraria from and is a consultant/advisory board member for Novartis. T. Price is a consultant/advisory board member for Amgen. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A. Townsend, N. Tebbutt, R. Joshi, T. Price

Development of methodology: A. Townsend, N. Tebbutt, C. Karapetis, T. Price

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Townsend, N. Tebbutt, C. Karapetis, N. Singhal, S. Yeend, L. Pirc, R. Joshi, J. Hardingham, T. Price

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Townsend, N. Tebbutt, C. Karapetis, N. Singhal, R. Joshi, J. Hardingham, T. Price

Writing, review, and/or revision of the manuscript: A. Townsend, N. Tebbutt, C. Karapetis, N. Singhal, R. Joshi, J. Hardingham, T. Price

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Townsend, P. Cooper, S. Yeend, L. Pirc, R. Joshi, T. Price

Study supervision: A. Townsend, N. Tebbutt, C. Karapetis, N. Singhal, R. Joshi, T. Price

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.