Novel Quality of Life Endpoints May Play Role in Guidelines

FORT LAUDERDALE--Evaluations of new treatments using traditionalendpoints such as response or survival may not be appropriatein advanced solid tumors that are highly symptomatic but essentiallyincurable, Howard A. Burris III, MD, said at the National ComprehensiveCancer Network conference.

Alternative quality of life endpoints may prove useful in evaluatingdrugs for these tumors, not only in clinical trials but also inguidelines development.

Dr. Burris and his group at the Cancer Therapy and Research Center,San Antonio, were involved in developing novel endpoints for studiesof gemcitabine (Gemzar) in advanced pancreatic cancer that ledto FDA approval of the agent.

Pancreatic cancer is highly lethal, with 86% of patients dyingwithin a year of diagnosis, Dr. Burris said. But a unique aspectof the disease is that almost all patients experience the samesymptoms--weight loss, jaundice, visceral pain, anor-exia, nausea,and depression. "Thus, it was an appropriate solid tumormodel to evaluate clinical benefit as an endpoint."

Previous discussions between NCI and FDA on using novel endpointspaved the way for the pivotal gemcitabine trials.

The San Antonio group's goal was to develop a "systematicand sensitive method to assess the palliative benefits of chemotherapy,"Dr. Burris said. They called this the clinical benefit response,measured as clinical benefit response rate. The gemcitabine studieswould also look at secondary endpoints such as tumor responserate, time to disease progression, and survival as a means ofvalidating the symptomatic endpoints.

Setting the Parameters

The primary measurement of clinical benefit response would bethe compilation of two factors--pain scales that measured analgesicconsumption and pain intensity to produce an overall pain evaluation,and traditional Karnofsky performance status (PS) measures.

Patients who were negative with regard to either of these measureswere by definition nonresponders. For those patients who werestable in these two categories, however, overall change in bodyweight was used as the tie breaker in determining whether thepatient would be counted as a clinical benefit responder.

Dr. Burris stressed that the improvements in these symptom parametershad to be significant for patients to qualify as responders. Avisual analog scale (0 to 100), developed with Dr. Russell Portenoyof Memorial Sloan-Kettering, was used to assess pain intensity,and patients had to have a 50% or more improvement over baselinethat was maintained for at least 4 weeks, to qualify as responders.

In addition, baseline pain had to be assessed at no lower than20 on the scale for patients to qualify as potentially benefitingwith regard to pain. "Those patients who had very littlepain with their disease going into the trial could not be positiveon this parameter," he said.

Analgesic consumption was measured by a daily diary kept by thepatient and a weekly count done in terms of morphine equivalentunits. Patients had to have a 50% or more improvement over baseline,maintained for at least 4 weeks, to be counted as positive responders;baseline consumption had to be at least 10 mg of morphine equivalentunits per day.

Two blinded independent evaluators were used to assess PS. Patientshad to have an improvement of at least 20 points over baseline,for at least 4 weeks, and baseline PS could be no higher than70 going into the trial.

Weight changes, used as a tie breaker, was "a very toughhurdle," Dr. Burris said. Patients had to have at least a7% gain in dry weight over baseline, maintained for at least 4weeks, to be considered positive responders; patients who developededema or ascites were excluded from the weight parameter.

Two trials of gemcitabine in patients with advanced pancreaticcancer were conducted using these endpoints. In the first trial,patients were randomized to weekly gemcitabine vs 5-fluorouracil(5-FU), with 63 patients on each arm.

Interestingly, he said, the literature proved true with regardto symptoms: 70% of patients had a PS of 70 or less; nearly 70%were taking 20 or more morphine equivalent units per day, andabout 70% had pain scores higher than 20.

In the gemcitabine arm, 15 patients had a positive improvementin pain vs 3 in the 5-FU arm. Four of the pain improvement patientsalso had PS improvement, and all of the patients with PS improvementalso had pain improvement.

In a subset analysis of clinical benefit responders vs nonresponders(15 on gemcitabine, 3 on 5-FU), median survival was greater than10 months for responders, twofold better than nonre-sponders,"and, more impressively, nearly half of the gemcitabine responderswere alive at 1 year," Dr. Burris said.

The Second-Line Trial

In the second-line trial in 63 patients who had failed 5-FU, gemcitabineprovided clinical benefit with regard to pain in 14 patients,and a few had improvement in both pain intensity and analgesicconsumption. Interestingly, Dr. Burris said, it was difficultto get patients off of analgesics, since investigators are inherentlyreluctant to dose reduce a patient who is not complaining of pain,"so reductions in analgesic consumption are clearly real."

Three patients had positive improvement in PS and were stablewith regard to pain, so 17 of 63 patients overall benefited fora 27% clinical benefit response rate (95% CI). Weight gain wasagain a "tough parameter," Dr. Burris said, with onlyone patient showing the required 7% improvement

Although tumor response rate was only 10.5% with gemcitabine,about 40% had stable disease or better as best response to gemcitabine.

Keep it Simple

In answer to a question from the audience, Dr. Burris said thathe thinks these parameters could potentially be incorporated intoan outcomes system for use in standard populations, to gauge theeffectiveness of chemotherapy in advanced cancer patients.

The challenge is to make the system simple enough for use in theclinic, he said. This is why other, more complex parameters, suchas hand grip strength and quality of life questionnaires, althoughused, were not primary endpoints in the gemcitabine studies.

"We left primary measurement as simple as possible--markinga line on a visual analog scale, counting pain medications taken,looking at PS as assessed by clinic nurses," he said. "Therewas no extra burden for physicians participating in the trial,which was really key to getting this trial done in a rapid fashion."

The 'KFC' Factor

Between 1991 and 1994, 24 new agents were tested for pancreaticcancer in the United States, and only two had a response rategreater than 10%, Dr. Howard Burris said at the NCCN conference(see story above).

In phase II studies of gem-citabine (Gemzar), the response ratewas only 11%, but "the interesting thing was that many pancreaticcancer patients had improvement in symptoms," Dr. Burrissaid. "Our director, Dr. Daniel Von Hoff, has called thisthe KFC factor, in that patients were out eating fried chickenand gaining weight while on study."

Gemcitabine was then selected for further study using novel endpointsto measure symptomatic response.