A man’s best friend is his PET

In this issue of Blood, Eertink et al¹ validate that a new radiomics-based prognostic classification outperforms the traditional International Prognostic Index (IPI) in identifying patients with diffuse large B-cell lymphoma (DLBCL) who are at high risk for treatment failure.

DLBCL is the most common lymphoma histology. Although this disorder is is an aggressive one, 50% to 70% of patients are cured with initial standard chemo-immunotherapy. However, currently no method is routinely available to identify, prior to therapy, those patients who are unlikely to benefit and should therefore be considered for an alternative treatment. The unfortunate consequence is that all patients with DLBCL are currently treated the same, regardless of differences in their predictable prognosis. At presentation, patients with DLBCL are given an anatomic stage, per the 4-stage Ann Arbor (AA) system that dates back to 1971. Next, they are assigned to a prognostic group according to the somewhat archaic, 30-year-old IPI, which uses simple clinical and laboratory features, including age, performance status, serum lactate dehydrogenase, number of extranodal sites, and AA stage. Unfortunately, neither AA stage nor IPI provides adequate information for therapeutic guidance. Thus, the range of treatment results has remained relatively stagnant.

Over the past 2 decades, the precision of staging and restaging has improved greatly, owing largely to the availability of 2^’-fluorodeoxyglucose positron emission tomography–computed tomography(FDG-PET-CT) scanning. PET-CT is more sensitive and specific than simple CT scans, and it helps distinguish viable tumor from fibrous tissue.² In several histologies, PET-CT has eliminated the need for subjecting patients to the dreaded bone marrow biopsy. Improvements in equipment and better standardization of interpretation with the 5-point Deauville score have further enhanced the usefulness of PET-CT. Such advances justified, in part, the revised staging and response criteria used to classify nodal lymphomas—the widely used Lugano classification of 2014.³ 

Recent enhancements in metabolic imaging have further improved the ability to predict, pretreatment, which patients are likely to benefit from therapy. Numerous studies have demonstrated that the quantification of metabolic tumor volume (MTV) derived from the PET-CT scan is highly correlated with patient outcome in DLBCL⁴ as well as other lymphoma histologies. Radiomics, or quantitative FDG-PET features, examines other characteristics of the lymphoma phenotype, including the peak standardized uptake value, the tumor shape and heterogeneity, and the greatest distance between the largest and most distant lesions.⁵ These, and many others, alone and in combinations, appear to outperform the IPI alone. Several groups have now developed prognostic systems combining one or more of these metabolic features with standard clinical features, producing promising results.⁶ 

In the current article, Eertink et al report the results of a validation study of the clinical PET scoring system, originally tested in the HOVON-84 study, now including a larger number of patients from 6 different clinical trials. Their system included MTV, the value for the greatest distance between the largest and most distant lesions, the peak standardized uptake value, and the simple clinical factors of patient age and World Health Organization performance status. The results achieved with clinical PET were superior to those achieved with the IPI, and perhaps other published radiomics-based systems, in distinguishing patients unlikely to do well with respect to 2-year progression-free survival, thereby offering a potential advance in the management of DLBCL patients (see figure). Unfortunately, although clinical PET improves the ability to identify high-risk patients by almost 10%, compared with the IPI, more than half of the group with the poorest prognosis (51.9%) were still free of progression or death at 2 years. Whereas the clinical PET data are quite encouraging, no group can be readily identified for whom the treatment is sufficiently unlikely to be favorable and should be altered de novo. Studies currently examining molecular genetic signatures⁷ and circulating tumor DNA⁸ provide further hope for identifying clinically meaningful patient subsets.

In June 2023, at the International Conference on Malignant Lymphoma (ICML)-17, in Lugano, Switzerland, a workshop will be convened to determine whether revisions of the Lugano classification for patient evaluation, staging, and response criteria for lymphomas are warranted.³ The large number of issues to be discussed include the following: Can we simplify, standardize, and improve upon the current anatomic staging system, incorporating prognostic factors, while making it more useful for a wide audience of physicians? Should MTV supplant CT as a measurement of tumor bulk? Are the newer technologies, such as MTV, radiomics, and circulating tumor DNA, ready for “prime time”? What is the role of these technologies in assessing minimal residual disease? The exciting preliminary data relating to use of these advances will make it hard to resist moving ahead with vigor to adopt them. However, we need to temper our enthusiasm a bit; for a new staging or prognostic classification to be useful, all the components must be not only validated but also widely available.

The holy grail for DLBCL is a risk-adapted approach in which the next generation of prognostic and predictive factors will help guide us in reducing treatment, and thus cost and toxicities, for the groups of patients for whom these methods are likely to be favorable, while altering our approach to improve outcome for those less likely to benefit from standard of care. Whereas clinical PET is clearly a major step in the proper direction toward individualization of therapy, there is still room for an upgrade. Incorporation of additional prognostic factors in the future should further enhance its performance. But, improved therapies and predictive biomarkers are needed to achieve true success. “Success is a science: if you have the conditions, you get the result” (Oscar Wilde).

Conflict-of-interest disclosure: The author declares no competing financial interests.