NCCN Is Pioneering Outcomes Research In Cancer Treatment

FORT LAUDERDALE, Fla--There is a void of outcomes data in cancerthat the National Comprehensive Cancer Network (NCCN) hopes tofill, said Jane Weeks, MD, at the NCCN's first annual conference.

Dr. Weeks pointed out that the Agency for Health Care Policy andResearch, the federal body responsible for performing outcomesresearch and developing practice guidelines, did not include asingle cancer diagnosis in its original 14 projects.

The NCCN, made up of 15 member institutions (see Table 1), hasmade a decision to invest in outcomes research, said Dr. Weeks,assistant professor of medicine, Harvard, and director of Dana-Farber'sCenter for Outcomes and Policy Research.

She called the network "an ideal source of data for outcomesstudies" for a number of reasons: the geographic variabilityof the member institutions and the subsequent variability in patternsof care; the variability of practices within the institutions,even within similar areas; and the innovative as well as standardapproaches being used in the NCCN member institutions.

She also cited the willingness of these institutions and theirclinicians to invest the resources necessary to collect and analyzeoutcomes data, and the critical mass of clinical experts and healthservices researchers within the member institutions.

To accomplish its goals for outcomes research (see Table 2), Dr.Weeks said, the NCCN began by taking an inventory of the existinginstitutional and national data sources and mechanisms of datacollection within the member institutions.

The data inventory project found that an enormous amount of datais currently being collected in the NCCN institutions, "butit was clear that data collection efforts by one person in a giveninstitution were often unfamiliar to other people collecting datain the same institution."

The survey also uncovered "enormous heterogeneity" inthe systems being used to collect data, she said. For example,among the 13 NCCN centers surveyed, there were 19 different vendorproducts for the hospital information systems, plus nine homegrown systems, "which makes it hard to pool data."

Dr. Weeks described a "narrow window of opportunity"to standardize systems before incompatible systems are in place."All the institutions are in the process of investing veryheavily in computer systems to collect outcomes data. It's a relativelytrivial thing now to ensure that the data are collected usingcommon definitions, but once those systems are in place, it isnot a trivial undertaking. So it's now or never on this, and Ithink we've agreed that it's going to be now."

On the bright side, the survey showed that the institutions arecommited to pooling data. "All institutions are now submittingtheir data and their forms centrally so as to generate some collectivedata," Dr. Weeks said. "The ultimate goal is to createa uniform outcomes measurement system to monitor guidelines compliance."A subset of the NCCN institutions is now engaged in putting togethera pilot project to monitor compliance with one or more of theguidelines developed by the NCCN committees.

Why Larynx Cancer?

One of the pilot efforts in gathering outcomes data to definepatterns of care is the NCCN Larynx Cancer Project, said DavidG. Pfister, MD, of Memorial Sloan-Kettering Cancer Center andCornell University Medical College.

"We have been asked why we chose to study larynx cancer ratherthan focus entirely on more common cancers," Dr. Pfistersaid. In approaching any outcomes research problems, he noted,there are basically two questions to ask: Is there practice variability,and, if so, is the clinical issue an important one?

In terms of larynx cancer, there is a very marked practice variabilityin patients with locally advanced disease, he said, with two basicapproaches for the primary site: total laryngectomy, generallywith postoperative radiotherapy; and larynx preservation throughone of three methods--partial laryngectomy, often with radiotherapy;radiotherapy alone; or radiotherapy with chemotherapy. In thelatter options, he said, total laryngec-tomy is reserved for salvagetherapy.

Is larynx cancer important to study? Dr. Pfister asked. "Itmay not be as common as lung, colorectal, breast, or prostatecancer, but within the upper aero-digestive tract, it is the mostcommon site of cancer. And treatment can have a profound effecton the patient's ability to communicate and quality of life."

He cited a Harvard study in which a significant number of peoplesaid they would be willing to accept a lower cure rate in orderto save their voice box.

Also, he said, there is a good randomized study of total laryngectomyand postoperative radiotherapy versus chemora-diation therapy,with total laryngectomy reserved for salvage, in patients withadvanced squamous cell larynx cancer (the Veterans AdministrationLarynx Preservation Study, published in 1991 in the New EnglandJournal of Medicine).

Now, with long-term follow-up, survival rates are quite similarand without statistically significant difference, he said, butin the chemoradiation arm, approximately two thirds of cured patientswere able to preserve their larynx.

In the larynx preservation arm, which used total laryngectomyas salvage therapy, T stage was predictive of ultimately savingor losing the larynx, he said. In patients with T4 disease, larynxpreservation was more likely to be ultimately unsuccessful thanin those with a less advanced T stage.

Because laryngectomy was the issue in this pilot project, theresearchers focused on advanced squamous cell larynx cancer. Patientshad to have resectable T2-T4 disease with no distant metastases,no prior cancer, and no prior treatment of their larynx cancer."We focused on patient accession from 1993 to 1994, sincethese patients were being treated approximately 2 to 3 years afterdissemination of the VA study results," Dr. Pfister said.

The seven participating institutions were each given a templatefor collecting their data and doing their analyses. For all seveninstitutions, 304 patients with T2-T4 disease were identified,illustrating some of the power that the NCCN has to address researchquestions requiring large numbers of patients, he said.

After the exclusions, 170 patients remained for analysis. "Wefound that larynx preservation was offered to 81% of T2 patients,70% of T3 patients, and 49% of T4 patients," Dr. Pfistersaid.

"I think this analysis demonstrates that the majority ofpatients were offered a larynx preservation approach, and thatthere was a correlation in this regard with T stage," hesaid. "Also, the denominator for our analysis was significantlyaffected by clinical and other patient selection factors, emphasizingthe importance of good communication between clinicians and outcomesinvestigators."

Dr. Pfister is encouraged that the institutions were able to successfullycollaborate to obtain the data for the larynx cancer project."We are in the process of obtaining and analyzing outcomesand benchmark data to provide a further perspective on this analysis,"he said.