Letter Expressing Concerns About Pending Approval of Cancer Drug Gefitinib (Iressa)
May 1, 2003
Janet Woodcock, M.D.
Director, Center for Drug Evaluation and Research
Food and Drug Administration
5600 Fishers Lane
Rockville MD, 20857
Dear Dr. Woodcock:
Public Citizen, representing 125,000 consumers nationwide, is concerned that the Food and Drug Administration (FDA) will grant approval to Iressa (gefitinib; AstraZeneca) when the deadline for this decision arrives on May 5, 2003. The sponsor is seeking accelerated approval (AA), but we believe that no class of approval is justified since Iressa appears to be both ineffective and dangerous. The pivotal trial on which AA for Iressa rests is a small uncontrolled, unblinded, Phase II trial in atypical non-small cell lung cancer (NSCLC) patients for third-line treatment. Of the co-primary endpoints, one (symptom improvement) could not be accurately evaluated and the other, a decrease in tumor size was reached in only 10% of patients. In addition, a well-conducted Phase III trial of Iressa as first-line therapy in NSCLC patients was unequivocally negative with respect to all endpoints. Based on our examination of the Iressa Briefing Documents (the background analysis done by the Oncology Division), the Oncology Division Advisory Committee (ODAC) transcript on Iressa, and the medical literature, we are concerned that there is not enough information to determine whether Iressa provides a benefit. In addition, as of February 2003, there were 173 post marketing reports of deaths from acute interstitial pneumonia associated with Iressa and apparently unrelated to NSCLC in Japanese patients (Japan is the only country to have approved Iressa).
AA is a category added to FDA regulations in 1992 to speed the process of getting new drugs to patients with diseases that are serious or life-threatening and for which there is no other treatment. Drugs approved for AA must be followed by a post marketing demonstration that the drug is “indeed associated with clinical benefit” employing randomized trials “usually in patients with less refractory tumors.” If such a Phase IV study fails to demonstrate clinical improvement, there is a process for rapidly removing the drug from the market.
In the case of Iressa, however, these randomized trial data (INTACT 1 and 2) are already available before possible approval and are “unambiguously negative.” According to the FDA Team Leader, “The FDA has never received a cancer drug application for accelerated approval when definitive data in another related setting show[ed] a lack of efficacy.” (Italics added) (See below).
Phase III Trials of Iressa as first-line therapy
First-line therapy is for patients who have not been exposed to prior therapy. Two such trials have been completed with Iressa, INTACT 1 and 2. According to the statistical reviewer these were, “well-conducted, double-blinded, placebo-controlled, randomized Phase III studies.” Together, these two studies included over 2,000 first-line NSCLC patients. (These are the studies that, under AA, would have been submitted during the post marketing period.)
Each trial had three arms: INTACT 1 tested gemcitabine plus cisplatin plus 250 mg of Iressa, gemcitabine plus cisplatin plus 500 mg of Iressa, and gemcitabine plus cisplatin plus placebo in a total of 1093 patients. INTACT 2 tested taxol plus carboplatin plus 250 mg of Iressa, taxol plus carboplatin plus 500 mg of Iressa, and taxol plus carboplatin plus placebo in a total of 1037 patients. The primary endpoint in both studies was overall survival.
Yet, in spite of the excellent design, the statistical reviewer concluded that, “there was no statistically significant difference between ZD1839 [Iressa] treated arms and placebo treated arm with respect to overall survival”; there was also no difference “with respect to secondary endpoints including response rate [tumor size] and time to progression” (italics added). This is in spite of the fact that such patients should have been the most responsive to treatment, since they have never been exposed to previous chemotherapy.
Pivotal Trial of Iressa as third-line therapy
The Phase II trial (Study #39) was submitted to FDA as the pivotal basis for approval for Iressa. In third-line treatment, patients have been exposed to at least 2 previous treatment regimens, in this case, both a platinum drug and docetaxel. These patients had failed first- and second-line treatments due either to unacceptable toxicity or disease progression while on therapy. Patients were randomly assigned to either 250 mg/day or 500 mg/day of Iressa. However, the Medical Officer considered this a single arm trial since it was not sized to compare the 2 arms and there was no other comparison group. Furthermore, both patients and caregivers were unblinded as to the response to therapy.
The sponsor specified two co-primary efficacy endpoints (objective tumor response and Lung Cancer Subscale [LCS], a symptom improvement rating scale).
Patients were required to have a baseline score of #24 (28 is asymptomatic). Although the sponsor considered a 2-point LCS improvement successful, members of the FDA review team questioned its meaningfulness. The team leader said that there was no way to know whether the 2-point improvement was due to the placebo effect (since there was no control arm), and there were no data validating LCS use in a single-arm study.
Both the Statistical Reviewer and the Medical Officer pointed out three major problems: 1) the lack of a control arm of patients not receiving Iressa, 2) the practice of telling patients who had an objective tumor response that their cancer had significantly decreased in size, which “can explain the correlation between response and symptom improvement”, and 3) providing most study patients with multiple concomitant medications such as drugs to treat “shortness of breath, cough, chest tightness, and ease of breathing” which could have “a profound effect on lung cancer symptoms that were evaluated.” Furthermore, no information on doses or schedules of these medications was collected, making it “impossible to determine whether ZD1839 [Iressa] or concomitant medications were responsible for the observed symptom improvement.”
Tumor response was based on tumor measurements and CT-scans. A responder was defined as someone who had a 50% decrease in the sum of the area of measurable lesions. The tumor response of the two dose groups averaged 10% (11.8% at 250 mg/day and 8.8% at 500 mg/day).
However, the FDA Medical Officer found these patients to be an atypical group in terms of disease severity and sites of disease. He stated that, “The responding patient population does not reflect a typical non-small cell lung cancer patient with metastatic disease.” The population was heavily represented by people with slowly growing, less aggressive tumors (66% had adenocarcinoma alone and 8% had adenocarcinoma mixed with squamous cell carcinoma - tumors that are the least aggressive and have the slowest doubling time of all lung cancer histologies). Indications of low biologic aggressiveness included a long median time from diagnosis to randomization (20 months) as well as the fact that most responding patients (16 out of 22) had a baseline status indicating none or only mild symptoms.
The FDA Medical Officer added that, “Since the large majority of patients enrolled in both trials [pivotal Study #39 and supportive Study #16] had stage IV disease, it might be expected that patients would have multiple sites of disease and, therefore, multiple measurable lesions. That was not the case."
The Medical Officer provided further background for his evaluation: “A widely accepted medical oncology principle is that for each chemotherapy regimen failed, the probability of responding to a subsequent regimen decreases.” As a result, one would expect the response rate in Study #39 (where patients were refractory to two or more prior chemotherapy regimens) to be lower than for patients who have had no prior treatment (INTACT 1 and 2). Again, “This was not the case.”
Adverse Event Information (clinical trial data)
There was one death in Study #39 due to pneumonia and one death due to acute respiratory distress syndrome (ARDS). In addition, ARDS and pneumonia occurred in 25% of patients (55 of 216) in Study #39 and in 36% of patients (39 of 107) in the supporting Study #16. Adverse event data from the Phase III have not been made public.
At the time of the ODAC meeting, no post-marketing data were available (the first deaths were announced in Japan three weeks after the advisory committee meeting); as a result, FDA was under the impression that Iressa was “relatively non-toxic.” Nevertheless, the team leader worried that small Phase 2 trials provided only limited safety data, and that the potential existed for new adverse events when, after approval, use would expand to a wider population. This appears to have happened in Japan.
Adverse Event Information (post marketing)
The only country to approve Iressa, thus far, is Japan (approved July 5, 2002). By October 15, 2002, there were reports of 13 deaths and by December 13, there were 124 deaths. By February 2003, AstraZeneca Japan announced 473 patients had developed acute lung disease and/or interstitial pneumonitis of whom 173 died. Since about 23,500 patients had received Iressa in Japan at that time, the death rate from lung disease associated with Iressa was 0.74%. Yet, this is almost certainly the lower bound since only a small proportion of cases are both properly diagnosed and reported.
Of a group of eighteen patients in Japan treated with Iressa, four contracted pneumonia and two died. The conclusion of the treating physicians, based on patient history and autopsy data, was that, “gefitinib caused the acute interstitial pneumonia because of previous alveolar damage from radiotherapy.”
Mechanism of action
Iressa completely blocks the epidermal growth factor receptor (EGFR). Since almost all cells have EGFRs, Iressa “not only inhibits the growth of cancer cells but also inhibits physiological replacement of normal cells, particularly after injury.” Many patients who take Iressa have previously received chemotherapy and/or radiation and are thus likely to have residual tissue damage, which they are unable to repair and which may, in turn, lead to pneumonia.
The Medical Officer stated that what one has in this case is pivotal trial that is a non-randomized, non-blinded, single-arm study in 139 patients with all patients essentially getting the same treatment. The tumor response did not conform to what was expected of first-, second- or third-line patients (patients lacked multiple lesions and response rate was unaffected by previous chemotherapy).
The Team Leader also questioned whether a 10% tumor response rate in 139 patients was likely to predict clinical benefit, especially since symptom improvement, the co-primary endpoint, failed in any meaningful sense and since the NSCLC patients were an atypical group with a non-aggressive form of lung cancer. Furthermore, the definitive large, post marketing studies that normally would have been the basis for approval of Iressa have already been completed and were completely negative in respect to longer survival.
The authors of an article in the International Society of Drug Bulletins suggested that based on their analysis of the Japanese Summary Basis of Approval for Iressa (in Japanese) and the increasing number of deaths due to interstitial pneumonia unrelated to NSCLC that, “the regulatory approval process should be completely re-examined.” From our reading of the available documents, we concur with that assessment.
We reach that conclusion with some reluctance. We understand that the gravity of NSCLC and the lack of effective therapy leave patients with few treatment options. But in this case, the data are either the product of poorly designed trials or, in the better-designed trials, unequivocally negative. Approval of any drug under these circumstances cannot be justified.
Elizabeth Barbehenn, Ph.D.
Peter Lurie, M.D., M.P.H.
Sidney M. Wolfe, M.D.
Public Citizen’s Health Research Group