Last week, the preliminary results of interim analysis the ESCAPE (Evaluation of Sorafenib, Carboplatin, And Paclitaxel Efficacy in NSCLC) trial were presented by Dr. Scagliotti at the 1st IASLC (International Association for the Study of Lung Cancer)-ESMO (Eurpean Society for Medical Oncology) Lung Cancer Conference in Geneva, Switzerland. This was a randomized, placebo-controlled, double-blind phase III study for patients who have not yet had chemotherapy for advanced NSCLC. Patients were treated with a standard course of chemotherapy (carboplatin and paclitaxel) and were randomized to receive sorafenib or placebo along with chemo. As you may recall from other posts, sorafenib (also known as nexavar) is an oral multitargeted tyrosine kinase inhibitor which primarily acts to block the VEGF receptors thus inhibiting tumour angiogenesis (blood vessel growth and development). Dr. West previously noted in a prior post that this trial was reportedly negative, but I was able to see the first presentation of the actual results.

In recent years the idea to combine antiangiogenesis agents with chemotherapy has become very popular, and this strategy was further encouraged by the positive results of the E4599 study that demonstrated a two month survival advantage when the monoclonal antibody to circulating VEGF, Avastin (bevacizumab), was combined with carboplatin/paclitaxel. Unfortunately, we have also recently learned that a similar trial with cisplatin/gemcitabine plus Avastin (AVAiL) failed to confirm this survival advantage (prior post here). What the future of this combination looks like is now unknown, at least for the countries who have not funded chemotherapy plus Avastin based on E4599.

In any case, the ESCAPE trial enrolled 926 patients over approximately 18 months (Feb 2006 – May 2007). The demographics of this population were very similar to other studies, about 62% male, medically fit for chemotherapy (ECOG performance status (PS) of 0 or 1), and no brain metastases at the time treatment started. Approximately 25% of patients had squamous cell carcinoma, the majority having adenocarcinoma: this is all very typical for phase III studies in advanced NSCLC. The toxicities were fairly predictable, and the median number of cycles of chemotherapy was 5 for the chemo alone group and 4 for the chemo plus sorafenib group. The response rates were also standard for the chemo alone arm with response rates of 5% complete (no evidence of residual cancer), 23% partial response, and 51% with stable disease. The chemo plus sorafenib arm was very similar with 1% complete, 30% partial response, and 46% with stable disease.

The progression-free survival was 5.1 months in the chemo alone arm and 5.4 months in the chemo plus sorafenib arm. The median survival was similar in both arms: 10.7 and 10.6 months in the control and experimental arms, respectively. Neither of these reached statistical significance. Thus, this trial did not meet its primary endpoint, which was to improve survival by the addition of sorafenib.

One interesting feature that came out of this study was the subgroup of patients (25%) who had squamous cell carcinoma. In this group the median survival for those who received sorafenib was worse than the control arm, 13.6 versus 9.8 months. In the non-squamous carcinoma group, the median survivals were similar at 10.3 and 11. 5months for the control and the sorafenib arm, respectively. The toxicities were not particular different for the patients with squamous cancers, though more patients died of progressive disease on the sorafenib arm (33%) than the chemo alone arm (25%). Overall, the reason this possible worse outcome for squamous patients on sorafenib plus chemo remains unknown -- however, the 13.6 month survival for the squamous arm is longer than one would expect from previous studies, so perhaps there was some biological imbalance that was not evened out by the randomization process.

The results of this study, although I stress that these are only the preliminary results, are troubling for a number of reasons. Of course it is disappointing that this strategy did not make a significance difference in survival. But more than this, I am concerned that this might be a trend, and this is particularly worrisome after the early closure of a very similar study (BR-24, run by the NCI-Canada) with carboplatin/paclitaxel with or without AZD2171 (cediranib, with the trade name Recentin), another oral multitargeted tyrosine kinase inhibitor of VEGF receptors. Although no results have been released from this study (it will likely be presented at the ESMO meeting in Sept 2008), it is troubling that it was stopped early after an interim analysis that showed it could not meet its primary endpoint to improve overall survival with the chemo-AZD2171 combination. There are rumours that that there were toxicity issues rather than a lack of efficacy, but still it is something of a disappointment.

The potential failure or problems with tyrosine kinase inhibitors of angiogenesis with chemotherapy reminds me of the earlier failures of the combinations of chemo and the oral EGFR inhibitors (INTACT 1 and 2, TRIBUTE, and TALENT; described in a prior post here). And now with the lack of improvement in survival in the AVAiL trail (cisplatin/gemcitabine with or without Avastin), perhaps lung cancer investigators, myself included, have to at least consider the possibility that this strategy isn’t working. Or it isn’t working for all types of NSCLC. Of course all of these studies have a solid foundation in basic (lab-based) research, with tumour models suggesting this is a successful strategy making the results in people all the more frustrating and disappointing.

To me, this reinforces the need to collect human tumour tissue and work closely with our colleagues in the labs to figure out some of the differences among various lung cancers. Perhaps it stimulates other questions out there.

As always, I’d be interested to hear your thoughts.