aGVHDAI: escape from the silo

Comment on Leisenring et al, page 749

A new dynamic tool for acute GVHD assessments opens a number of possibilities for use in observational and interventional clinical trials and in clinical care.

The “silo effect” is a term for a lack of communication between components of an organization. The medical scientific community is not immune to such patterns of behavior. The article in this issue of Blood by Leisenring and colleagues is an example of how an open-minded adoption of concepts from one medical specialty can contribute to another area of medicine, in this case to effectively address the long-standing quandary of how to develop better tools to assess acute graft-versus-host disease (GVHD).^1-4  The Acute GVHD Activity Index (aGVHDAI) described here was inspired by the Crohn Disease Activity Index, a system for scoring the severity of Crohn disease.⁵  Classic acute GVHD grading systems based on arbitrary categorical cutoff points of skin rash, diarrhea volume, and bilirubin levels have established prognostic value but are insufficiently accurate to predict survival, indicating the possible importance of using different additional clinical variables. Clinical trials that use such grading systems may overestimate or underestimate the benefit of a given therapeutic intervention. Prognosis in existing grading systems is highly dependent on the peak score value and is therefore impractical for real-time use in making clinical decisions. Grading systems should also be relatively easy to use, which is certainly not the case with the current system.

Leisenring and colleagues address some of these deficiencies by developing a new method for more accurately predicting nonrelapse mortality in acute GVHD patients by using 10-day observation interval data from 386 patients with acute GVHD following matched unrelated donor bone marrow transplantation for chronic myelogenous leukemia (CML) (see figure). By using statistical modeling of several clinical variables, the authors identified the optimal prediction model (aGVHDAI), which included total serum bilirubin, markedly decreased oral intake, need for systemic treatment, and Eastern Cooperative Oncology Group (ECOG) performance score. Unlike peak grading, which is static, this method incorporates the dynamic concepts of time and ongoing need for therapy. There are 2 proposed primary uses of this index: (1) to assess the risk of fatal outcome in real time, thereby aiding therapeutic decisions and stratifications in trials; and (2) to provide an accurate research tool for measuring the burden of acute GVHD across time as a primary outcome in prevention or treatment studies.FIG1 

The authors should be commended for this refreshing effort to improve acute GVHD classifications. However, many questions remain about aGVHDAI, and much work remains to be done before this index can be recommended for routine use. Although aGVHDAI eliminates the need for measuring diarrhea and skin surface area, it reintroduces performance status and adds calorie intake as variables that could pose new reproducibility challenges. Also, it will be interesting to assess the prognostic value of aGVHDAI for predicting other relevant clinical outcomes, such as treatment failure and overall survival. This study is retrospective; aGVHDAI will need to be evaluated in patient cohorts studied prospectively in order to assess the index's real-time utility. Finally, the new system needs to be validated for its accuracy and reproducibility in other transplantation centers, in diseases other than CML, and in different transplantation patient populations, such as children, recipients of related-donor transplants or peripheral blood allografts, or in patients treated in reduced-intensity conditioning settings. ▪