The need for personalized information at the point of care

In a recent issue of the Journal of Clinical Oncology, Joseph Connors, MD, writes another nice editorial that reflects on the issue of initial treatment intensity in advanced-stage Hodgkin’s lymphoma.

For many of us, doxorubicin, bleomycin, vinblastine and dacarbazine (ABVD) combination chemotherapy remains the mainstay of initial treatment for almost all newly diagnosed advanced-stage Hodgkin’s lymphoma presentations.

However, several investigators — including the German Hodgkin Study Group (GHSG) — have explored strategies such as escalated BEACOPP to increase treatment intensity to reduce relapse rates and improve OS.

Connors discussed the GHSG’s recently completed HD12 trial and noted the confirmation of previously described impressive cure rates (85%) with this regimen.

Acknowledging that this regimen may improve freedom from primary treatment failure in comparison with regimens such as ABVD, Connors nonetheless used another recently completed trial by the Intergruppo Italiano Linfomi to illustrate that the effects on long-term OS — the endpoint that should matter — are less clear.

William Wood

Indeed, the Intergruppo Italiano Linfomi trial showed a 10-year OS with ABVD (87%) that essentially was the same as the OS in the GHSG HD12 trial using escalated BEACOPP.

The escalated BEACOPP regimen comes with a price — 3% treatment-related mortality, 2% to 3% leukemogenesis, near-universal sterility in men and high health care utilization — beyond what would be expected with ABVD.

So, which regimen is best? The issue is complicated further when one considers that there are several other regimens besides these two from which clinicians can choose when treating newly diagnosed young patients with Hodgkin’s lymphoma.

How can a clinician choose which regimen to use for a given patient at the point of care? Because of the challenges in comparing published trial data — or even the challenges of knowing the published trial data — cognitive biases can prevail.

I have some colleagues who I would consider more aggressive in their approach to treatment who always prefer escalated BEACOPP for high-risk advanced presentations. I have others who I would consider more conservative who would not. But should cognitive bias be the deciding factor?

I recently attended a geriatric oncology retreat hosted by my institution, the University of North Carolina at Chapel Hill. The speakers were very good, and they included Heidi Diana Klepin, MD, of Wake Forest Baptist Health.

Klepin discussed the challenges of treating older adults with acute myelogenous leukemia and noted the potentially powerful prognostic information imparted by patient-specific risk factors such as performance status and physical fitness.

At the same time, our current measures of performance status, such as the ECOG scale, clearly are inadequate in differentiating the “marginally fit” older patient population to determine who is at the greatest risk for induction therapy-related toxicity.

To address these problems, Klepin has begun to incorporate comprehensive geriatric assessments (CGAs) into her risk stratification approach for the newly diagnosed elderly with AML. Klepin’s data suggest that CGAs are prognostic in this patient population; in fact, the Cancer and Leukemia Group B (CALGB) has now incorporated CGAs into two of its elderly AML therapeutic trials.

Further, Klepin has begun work on non-pharmacologic interventional trials, such as moderate exercise, to improve the outcomes of vulnerable patients.

Obviously, I hope Dr. Klepin’s work is successful. Unless the CGA becomes part of standard assessments for elderly AML patients, though, what will we do with the new prognostic information that we have learned? How will a clinician integrate biologic data and functional data into an AML management plan for a specific patient?

In current practice, Klepin said, decisions are highly variable. Analogous to the Hodgkin’s case, there are some centers that are both “traditional” and “intensive,” treating nearly all older AML patients with standard 7+3-like induction therapy.

There are other centers that are more “early adopters” and “less intensive” and treat nearly all older AML patients with hypomethylators or combinations of these drugs with other new agents.

Which is right, and should geography be the deciding factor?

It seems to me that we need more effective ways of translating our accumulated wealth of experimental and observational data into clinical decision support at the point of care.

I, for one, would love a platform that allowed me to input data about a specific patient’s disease biology, host genetic factors, clinical comorbidities and functional status, and receive in return projected personalized survival estimates and expected short- and long-term toxicity profiles for competing treatment strategies based on the aggregate of the world’s known data about those who have received these treatments for these diseases.

Of course, the requirements for amassing that kind of data and constructing that kind of database would be astronomical, and I know many would point out the statistical pitfalls and flaws in constructing those kinds of models.

I would counter: Is cognitive bias or geography a better way to decide? Although the investment will be substantial and the short-term payoffs difficult to realize, we owe it to our patients to construct models that enhance the delivery of personalized information and treatment planning at the point of care.

William Wood, MD, is assistant professor of medicine in the division of hematology/oncology at the University of North Carolina in Chapel Hill. He may be reached at william_wood@med.unc.edu. Disclosure: Dr. Wood reports no relevant financial disclosures.

For more information:

• Borchmann P. J Clin Oncol. 2011;29:4234-4242.
• Connors JM. J Clin Oncol. 2011;29:4215-4216.
• Wood WA. Blood. 2011;117:5803-5815.