Access to AlloTx for AML: the value of the social phenotype Free

In this issue of Blood, Wuliji and colleagues evaluate the relationship between social determinants of health and access to hematopoietic transplantation in management of adults with acute myeloid leukemia (AML).¹ 

Since the beginning of this century, clinical features of hematologic malignancies, such as age, sex, history of antecedent hematologic disorder, as well as disease-specific clinical characteristics, have been enhanced, and often surpassed, by biologic features, such as cytogenetic and molecular changes that confer prognostic value. Defining these biologic features has given insight into how clonal hematologic neoplasia develops as well as how disease features may inform the development and utility of novel therapeutics. Less appreciated is how clinical characteristics, in addition to emerging social health determinants, may impact access to costly novel and not-so-novel therapeutics within the spectrum of diseases treated by hematologists.

Investigators from several academic medical centers prospectively evaluated the relationship between clinical characteristics, specifically what today is termed social determinants of health (SDOH), and access to hematopoietic transplantation in the management of adults with AML. The background consists of the hypothesis that SDOH can reduce access to allogeneic transplantation, as seen with other treatments.^2-4 The authors cite limited access to transplantation centers as well as impediments, such as provider bias and racial inequities, that may restrict availability of this potentially life-saving resource. For good or ill, the latter term is triggering, requiring definitions of both race and inequity as it applies to a medical technology.^2,5 Multiple barriers exist for hematopoietic transplant, a dose-intensive chemo-immunotherapy intervention limited by patient characteristics as well as by organizational constraints, such as expected registry outcomes, and the worrisome impact of outcomes on institutional performance metrics that threaten to limit access based on provider contracts. The results of such metrics also inform patient selection at the local level.⁶ Selection metrics could have been measured by the authors, such as the likelihood that a patient with a set of disease characteristics may be rejected based solely on a social determinant. How the authors would have accessed that information is uncertain. Instead, they had little choice but to analyze educational and occupational barriers, availability of care, housing insecurity and financial barriers, and the impact of institutional selection based on those barriers, rather than referral bias, which is often hidden and less likely to be available for analysis.

Notwithstanding those limitations, the authors strive to get at what limits access to intensive therapy by reporting patient data from 13 collaborating centers in the United States. A major limitation of the analysis is that only 2 of these centers would be expected to have a significant population of Hispanic patients, and the study was restricted to those patients who could speak and read English. A reliance on English language fluency, dictated by the survey instruments used, affected the ethnic heterogeneity of patient recruitment. One may think this reliance would not be required given artificial language processing and translational tools presently available.

A major strength of the analysis is that a large majority of potentially eligible subjects at the participating institutions gave written informed consent to participate, and few potential participants were excluded a priori and not offered participation.

With the available data, the authors described the likelihood that a patient would undergo allotransplant. Despite the small sample size, transplant was less likely among Black patients when compared with White patients, and Asian patients were more likely to undergo allogeneic cell therapy. The absolute percentages of patients from underrepresented minority groups receiving allotransplant did not differ greatly from the percentages reported by the Center for International Bone Marrow Transplant Registry. The authors state that clinical variables and AML subtypes did not vary sufficiently among the ethnic groups to account for the differences predicted by SDOH; however, given the small sample size, those clinical and disease-specific variables were not independently predictive of proceeding to allotransplant.

SDOH also predicted for outcomes after allotransplant after the impact of those factors had been adjusted for various clinical- and disease-specific features. Thus, it is fair to assume that the difference in outcomes is more likely related to a unique set of societal measures. Analyzing causes of death after allogeneic hematopoietic cell transplantation, however, is often limited by: 1) lack of unified definitions of different causes of death and 2) low numbers of patients. The authors looked at differences in relapse vs nonrelapse causes of death among educational, housing, and financial factors, and identified a relationship between these latter variables and their impact on increasing death from nonrelapse causes rather than relapse, with limitations imposed by sample size.

The greatest power to detect difference was the highest level of education achieved and use of Supplemental Security Income (SSI) in the zip code area analyzed. What is not clear is the impact of family size and involvement in health care for those areas with the least achievement of high education level and the greatest use of SSI. Also uncertain is what sample size, and what forms of data collection, would be required to answer the question of family involvement. There are still other questions regarding diverse underrepresented populations and their unique home environment.

Thirty years ago, it was not unusual for medical practitioners in hematologic oncology to make home visits, especially near the end of life. It was often observed that 5 minutes in a patient’s home provided insight into unique challenges that many months of clinic visits could not elicit. Now that home visits and assessments are either not done at all or referred to other health-care personnel, it is incumbent on our community to use an accessible, portable, and detailed online tool that would provide the health-care team with an assessment of what I term the patient’s unique “social phenotype.”⁷ This instrument should be made available to patients, regardless of language, and include, at a minimum, the domains listed in the table. Whether results are then calculated into a score to be prospectively analyzed will be the work of future studies by teams like the one assembled by Wuliji and colleagues.

Conflict-of-interest disclosure: G.S. declares no competing financial interests.