Unbiased Analysis of Today's Healthcare Issues

Q-TWiST: One approach for measuring quality-adjust survival gains in cancer

Written By: Jason Shafrin - Feb• 27•18

Defining which cancer treatments are the “best” is not as easy as you think.  At first glance, treatments that produce the largest survival gains should be considered best.  Most cancer treatments come with serious side effects.  Thus, new treatments that better improve survival may also increase the risk of adverse events (although in some cases–like the case of immuno-oncology treatments–adverse events appear less frequently than standard of care).

How can we evaluate the tradeoff’s between life extension and quality of life?   On standard approach for cost-effectiveness analysis is using quality-adjusted life years (QALYs).  This approach measures gains in survival improvements but does not value all survival equally.  One year survival where patients are in perfect health is valued at 1.0, a 1 year survival gain in less than perfect health is valued at less than 1.0.  For instance, an additional QALY of 0.5 would imply that the person was indifferent between living half a year at perfect health and living for 1 year at this level of less than perfect health.

Another approach for measuring the “quality” of any survival improvements is measuring time spent without symptoms of disease or toxicity of treatment (TWiST).  In a paper by Gelber et al. (1996), the authors use a 3 step method to measure Q-TWiST of postoperative radiation therapy alone or radiation therapy plus fluorouracil-based chemotherapy among patients with poor prognosis, resectable rectal cancer.  These steps include:

  1. Defining quality-of-life-oriented end points
  2. Partitioning overall survival time
  3. Comparing the treatment groups with respect to quality-adjusted survival

In the first step in the Gelber paper, they define TWiST as time period without: any adverse event during or soon after chemotherapy or radiation treatment, any subsequent delayed adverse event after treatment, or any disease relapse.  Next, the authors needed to allocate, what time should be considered as in suboptimal health when these adverse events occur.  In the Gelber paper, they assume that all time during treatment, time when adverse events were reported plus some additional time in case follow-up reporting was not completed (e.g., 2 extra days for nausea, 10 extra days for any other adverse event).  All time during relapse was considered suboptimal.

Based on these assumptions, the authors could calculate a quality-adjusted survival relative to TWiST (Q-TWiST) as follows:
Q-TWiST =TWiST + (utox × TOX) + (urel × REL)

Where TOX represents time spent addressing symptoms of adverse events and REL represents the survival time following a diagnosis of relapse.  The parameters utox and urel are the utilities when patients have toxic adverse events and disease relapse respectively. By assumption, time spent in TWiST is assumed to have a utility level of 1.o (i.e., full health)

For each of the two treatments considered–radiation alone vs. radiation + chemotherapy–the authors estimated Kaplan—Meier curves for time in TOX and disease-free survival in order to used to partition the overall survival time.  The authors then used these curves to estimate the amount of time spent in TWiST, TOX and REL.

Using this approach, the authors found that the combined radiation and chemotherapy treatment resulted in

more TWiST (6.1 months; 95% CI = 0.2-12.0 months) than patients who received adjuvant radiation therapy alone. Despite an increase in the amount of time that individuals spent with early and late toxic effects, the Q-TWiST analysis indicated that the combined therapy conferred significantly greater benefit for a wide range of patient preferences about living with the toxicity of treatment or the symptoms of overt disease.


You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

Leave a Reply

Your email address will not be published. Required fields are marked *