What Thresholds Should Be Used in Risk Prediction Statistical Models for Advanced Hodgkin Lymphoma?

In a research letter in Blood Advances, first author Matthew J Maurer, ScD, MS, of Mayo Clinic in Rochester, Minnesota, and international colleagues discussed applications of the Advanced-stage Hodgkin lymphoma International Prognostic Index (A-HIPI) model.

According to the letter, binary prognostic models in classical Hodgkin lymphoma, such as the International Prognostic Score, are easy for clinicians to use but at the cost of prediction accuracy. Thanks to advances in statistical software and the widespread availability of smartphones and tablets, more sophisticated models, such as A-HIPI, can now be used to make predictions for individual patients in a clinician’s practice. These individual patient risk assessments can also help establish the statistical thresholds for risk groups in clinical trial designs.

However, Maurer and colleagues explain that clinicians need guidance on using individual patient risk assessments to inform the risk group structure they would use in their practices or set for a clinical trial. They write specifically about the A-HIPI model, noting that, though it “provides a rigorous, accurate, and individualized estimate of risk for patients with AS-HL [advanced-stage classic Hodgkin lymphoma], there are challenges with defining risk groups from individual risk predictions.”

The coauthors then discussed the strengths and weaknesses of the A-HIPI model in defining different risk groups based on risk calculated for individual patients. A particular issue they observed is the asymmetrical distribution of 5-year progression-free survival (PFS) and 5-year overall survival predictions produced by the model.

In the A-HIPI model, the coauthors explored the statistical analysis benefits and drawbacks of increasing or decreasing the risk percentiles used to define low versus high-risk disease. Making such adjustments is complicated because clinicians may differ in the thresholds they regard as defining low- versus high-risk disease. The coauthors encountered this difficulty even when they inquired with AS-HL experts from the global HoLISTIC Consortium, the developer of the A-HIPI model.

“Although this approach leverages expert clinical judgment from those treating AS-HL and primary users of the model, the right-skewed distribution of A-HIPI risk scores greatly limited the applicability of this approach. Proposed cutoffs of PFS5 [5-year PFS] <70 and PFS >90 only identified 15% and <1% of the population, respectively,” Maurer and colleagues elaborated.

From their examination of the A-HIPI model, Maurer and colleagues found that “[a] flexible rank-based approach appeared to provide the most clinical utility and data granularity, which may be harnessed for future AS-HL clinical trial design and patient-stratification need.” They add, however, that further investigation is needed to confirm the proposed thresholds used to define low- and high-risk disease.

As clinicians use A-HIPI to design a clinical trial in AS-HL, Maurer and colleagues advise that “[d]efining a high-risk A-HIPI cutoff will depend on the needs of the study and/or user,” and “[s]ample size (availability and cost), event rate (power), and study goals should be weighed when selecting an optimum cutoff for an individual clinical study.”

Reference

Maurer MJ, Parsons SK, Upshaw JN, et al. The A-HIPI prediction model in advanced-stage Hodgkin lymphoma: identification of risk groups and creation of an online tool. Blood Adv. 2025;9(6):1366-1369. doi:10.1182/bloodadvances.2024014689