Cancer Research Lab

“Conditional survival” is a more accurate way to portray prognosis for cancer patients than is a typical actuarial table, physicians report in the Journal of Thoracic Oncology for March.

Historically, survival has been estimated based on risk at diagnosis, Dr. Samuel J. Wang and his associates note, whereas conditional survival takes into account patients’ declining risk over time.

So if a patient is still alive 3 years after diagnosis, his chances of continued survival are better than they were during the first year, Dr. Wang, from the Oregon Health and Science University in Portland, and his associates explain.

The researchers’ analyses are based on the Surveillance, Epidemiology, and End Results (SEER) Program from the National Cancer Institute.

The investigators’ target was patients diagnosed between 1988 and 1995 with non-small-cell lung cancer, followed through 2002. Their goal was to compute 5-year conditional survival rates for patients who had already survived up to 5 years after diagnosis.

The 96,480-patient cohort was 80% white, 4% Hispanic, 12% African-American, 5% Asian or Pacific Islander, and 0.4% Native American or Alaskan Natives. Other model variables were age (greater than or less than 70 years), gender, and stage of cancer. The authors presented their results as a set of histograms.

Survival increased further each year after diagnosis. Overall 5-year conditional survival ranged between 10% and 20% during the first year, to about 70% at year 5.

Generally, ethnic groups were more alike than different, Dr. Wang said, adding that they tended to “huddle together” as their “conditional survival curves moved in tandem.”

In most conditions, African-Americans had slightly lower relative conditional survival, but Dr. Wang noted that the difference rarely rose above 5% compared with whites. The discrepancy was widest “in earlier stage disease, male patients, age > 70 years, and squamous cell histology.”

In contrast, for blacks with “advanced stage disease or large cell histology or who were female,” conditional survival increased more over time, exceeding that of white individuals.

Stage I and II cancers did not change much over time, but stages III and IV saw much steeper increases in conditional survival over time.

“Our next step will be to use more sophisticated statistical analysis to generate a more accurate predictive model,” Dr. Wang said. As a result, they hope to identify “better parameters that would allow us to make individualized predictions for patients.”