Panitumumab and irinotecan versus irinotecan alone for patients with KRAS wild-type, fluorouracil-resistant advanced colorectal cancer (PICCOLO): a prospectively stratified randomised trial

Summary Background Therapeutic antibodies targeting EGFR have activity in advanced colorectal cancer, but results from clinical trials are inconsistent and the population in which most benefit is derived is uncertain. Our aim was to assess the addition of panitumumab to irinotecan in pretreated advanced colorectal cancer. Methods In this open-label, randomised trial, we enrolled patients who had advanced colorectal cancer progressing after fluoropyrimidine treatment with or without oxaliplatin from 60 centres in the UK. From December, 2006 until June, 2008, molecularly unselected patients were recruited to a three-arm design including irinotecan (control), irinotecan plus ciclosporin, and irinotecan plus panitumumab (IrPan) groups. From June 10, 2008, in response to new data, the trial was amended to a prospectively stratified design, restricting panitumumab randomisation to patients with KRAS wild-type tumours; the results of the comparison between the irinotcan and IrPan groups are reported here. We used a computer-generated randomisation sequence (stratified by previous EGFR targeted therapy and then minimised by centre, WHO performance status, previous oxaliplatin, previous bevacizumab, previous dose modifications, and best previous response) to randomly allocate patients to either irinotecan or IrPan. Patients in both groups received 350 mg/m2 intravenous irinotecan every 3 weeks (300 mg/m2 if aged ≥70 years or a performance status of 2); patients in the IrPan group also received intravenous panitumumab 9 mg/kg every 3 weeks. The primary endpoint was overall survival in KRAS wild-type patients who had not received previous EGFR targeted therapy, analysed by intention to treat. Tumour DNA was pyrosequenced for KRASc.146, BRAF, NRAS, and PIK3CA mutations, and predefined molecular subgroups were analysed for interaction with the effect of panitumumab. This study is registered, number ISRCTN93248876. Results Between Dec 4, 2006, and Aug 31, 2010, 1198 patients were enrolled, of whom 460 were included in the primary population of patients with KRASc.12–13,61 wild-type tumours and no previous EGFR targeted therapy. 230 patients were randomly allocated to irinotecan and 230 to IrPan. There was no difference in overall survival between groups (HR 1·01, 95% CI 0·83–1·23; p=0·91), but individuals in the IrPan group had longer progression-free survival (0·78, 0·64–0·95; p=0·015) and a greater number of responses (79 [34%] patients vs 27 [12%]; p<0·0001) than did individuals in the irinotecan group. Grade 3 or worse diarrhoea (64 [29%] of 219 patients vs 39 [18%] of 218 patients), skin toxicity (41 [19%] vs none), lethargy (45 [21]% vs 24 [11%]), infection (42 [19%] vs 22 [10%]) and haematological toxicity (48 [22%] vs 27 [12%]) were reported more commonly in the IrPan group than in the irinotecan group. We recorded five treatment-related deaths, two in the IrPan group and three in the irinotecan group. Interpretation Adding panitumumab to irinotecan did not improve the overall survival of patients with wild-type KRAS tumours. Further refinement of molecular selection is needed for substantial benefits to be derived from EGFR targeting agents. Funding Cancer Research UK, Amgen Inc.


2) Laboratory Methods
Formalin-fixed, paraffin-embedded tumor tissue was retrieved, anonymized and encoded with the patient's trial number at the treating hospital, then sent to the research laboratory. Here, all staff were blind to the patients' treatment allocation and clinical outcomes. The laboratory processing and storage of samples was under Ethical approval, and the laboratory adheres where possible to GCLP guidelines and participates annually in the UK National External Quality Assessment Service (NEQAS) scheme.
Areas containing the highest density of tumour cells were identified on a hematoxylin and eosin stained section. Six to nine 5 µm sections were used per extraction, depending upon the tumour area per slide. The sections were macrodissected with a scalpel blade, to ensure that only the tumour-rich areas were used. DNA was extracted using the Qiagen QiaAmp Micro Kit (Qiagen, Crawley, UK) following the manufacturer's instructions. The final DNA was eluted into 25 microlitres of laboratory-grade water and the concentration determined using a Nanodrop ND-1000 spectrophotometer (Labtech International, Uckfield, East Sussex, UK). The DNA was stored at 4 o C until required.
Primers for PCR amplification and Pyrosequencing analysis were designed using proprietary Pyrosequencing assay design software version 2.0.1.15 (Qiagen, Crawley, UK). PCR reactions contained 12.5µl of Qiagen HotStarTaq Master Mix (Qiagen, Crawley, UK), additional magnesium chloride to a final concentration of 2mM, 200nM each of forward and reverse primers, 20ng of tumour DNA and sufficient water to make a final volume of 25l. Thermal cycling conditions for all amplicons were 94°C for 12 minutes followed by 40 cycles of 94°C for 10 seconds, 55°C for 20 seconds and 72°C for 20 seconds. PCR products were sequenced by Pyrosequencing on a PyroMark ID system (Qiagen, Crawley, UK) following the manufacturer's protocols. Data was analysed by visual inspection of Pyrograms and by statistical analysis of peak height data.

Figure 1 (appendix): Subgroup analysis by mutation status: Response Rate
RECIST response rate by treatment arm in patients in the primary analysis population. "All-wt" = no mutations detected; "Any-mut" = any mutation detected.  Figure 5A-D

4) Sensitivity analyses a) All-mut status definition sensitivity analysis
In the main analysis, patients with an incomplete set of genetic data are included in the "All-wt" group provided their status is wt at every codon for which data is available. In Table 4 (Webappendix), these data are shown in the top row, and for comparison, in the second row, the data for patients with a full set of data confirming wildtype status at all 12 loci under investigation (n=237). Data are given for each efficacy endpoint for this patient group and the tests for interaction with "any-mut" status. In the main analysis, mutations at PIK3CA exon 9 or 20 are grouped along with mutations in the RAS-RAF-MEK pathway for inclusion in the "Any-mut" group, shown here in the second row (n=137) of Table 5 (Webappendix). However the role of PIK3CA in EGFR signalling is less direct. The sensitivity analysis presented here shows treatment impact when the 26 patients with only a PIK3CA mutation are not included (third row, n=111). Also shown are the treatment effects in patients with only a PIK3CA mutation, although numbers here are extremely small, and the confidence intervals correspondingly wide.