The past 20 years have witnessed an evolution in the assessment of response in lymphomas. The International Working Group published the first standardized response criteria in 1999, based on CT scans and other suboptimal technologies. The entry of PET and PET/CT into routine clinical practice led to a revision of those criteria in 2007; however, because they were PET-based, they were primarily used for diffuse large B-cell lymphoma and Hodgkin lymphoma, histologies that represented most of the PET experience. Increasing data with PET in other lymphomas and the development of standardized interpretation criteria — the Deauville Criteria — provided the stimulus for the so-called Lugano Classification, now 5 years old. These criteria considered PET/CT the standard method for end-of-treatment restaging, as well as for interim evaluation. Patients with a residual mass could still be considered to have a complete response if the mass was no longer fluorodeoxyglucose, or FDG, avid.
The study by Stephens and colleagues represents 5-year follow-up of S0816, a risk-adapted approach to the treatment of patients with advanced Hodgkin lymphoma. Although the OS rate remained excellent at 94%, 25% of patients with advanced disease relapsed despite a negative interim PET. The explanation for that observation was ambiguous to the investigators, and the only pretreatment characteristic associated with this finding was race. One proposed possibility is that there were differences between local interpretation of scans and central review.
My conclusion is that, once again, current response criteria need to be revised. Firstly, newer agents may produce unanticipated findings. The most striking are the flare reactions noted on PET/CT scans with immunomodulatory agents such as, but not limited to, checkpoint inhibitors. A modification of the Lugano Classification, called the LYRIC criteria, accounted for this observation.
However, other improvements in the current response evaluations need to be considered to improve on their predictability. Currently, we use the 5-point Deauville scale to compare the standardized uptake value (SUV) of the tumor with a background source, such as the liver. Recent data suggest better predictability with interim as well as end-of-treatment PET/CT using the delta SUV, the extent to which the SUV decreases in a relative fashion rather than absolute Deauville score.
More importantly, PET/CT do not consider the biology of the tumor and its microenvironment. Several investigators have suggested that tumor- and milieu-associated biomarkers may complement PET/CT findings, dividing patients with a negative scan into various prognostic groups.
Finally, PET/CT scans have a limit to their resolution, and the importance of minimal residual disease (MRD) appears increasingly compelling. This point is exemplified by two large randomized trials in follicular lymphoma, GADOLIN and GALLIUM. The former compared bendamustine plus obinutuzumab (Gazyva, Genentech) with bendamustine alone in rituximab (Rituxan; Genentech, Biogen)-refractory patients. The latter compared chemotherapy with either rituximab or obinutuzumab. In both of these trials, response rates were similar between the two arms, yet there was a PFS improvement in GALLIUM, and both an extension of PFS as well as OS in GADOLIN. The explanation was a difference in the ability of the superior regimen to eradicate MRD. Other studies have demonstrated that the so-called liquid biopsy may be more sensitive than a PET scan in identifying early relapse.
To be a bit blunt, perhaps we are doing this whole treatment thing wrong. It is possible, prior to treatment, to distinguish patients who are likely to benefit from those who will not. For example, total metabolic tumor volume or tumor heterogeneity prior to treatment appear to be powerful predictors of outcome in Hodgkin lymphoma and non-Hodgkin lymphoma. Several investigators have performed genetic analyses on patients with DLBCL and identified subsets with a likelihood of benefit from standard therapy ranging from less than 10% to 80%. Yet, we still treat all patients the same.
What is needed is an anticipatory risk-adapted strategy in which patients who are likely to do well receive standard treatment. Those unlikely to benefit are spared the wasted time and toxicity and are placed on clinical trials with novel approaches.
As Sherlock Holmes said, one must “never trust to general impressions, my boy, but concentrate yourself upon details.” A clearer understanding of the details will certainly result in improved patient outcomes.
Agostinelli C, et al. Lancet Haematol. 2016;doi:10.1016/S2352-3026(16)30108-9.
Cheson BD, et al. Blood. 1996;87(12):4990-4997.
Cheson BD, et al. Blood. 2016;doi:10.1182/blood-2016-05-718528.
Cheson BD, et al. J Clin Oncol. 2007;25(5):579-586.
Cheson BD, et al. J Clin Oncol. 2014;doi:10.1200/JCO.2013.54.8800.
Chapuy B, et al. Nat Med. 2018;doi:10.1038/s41591-018-0016-8.
Cottereau AS, et al. Blood. 2018;doi:10.1182/blood-2017-07-795476.
Pott C., et al. Leukemia. 2019;doi:10.1038/s41375-019-0559-9.
Schmitz R, et al. N Engl J Med. 2018;doi:10.1056/NEJMoa1801445.
Bruce D. Cheson, MD
Georgetown Lombardi Comprehensive Cancer Center
Disclosures: Cheson reports consultant roles with AbbVie, AstraZeneca, Celgene, Gilead, Morphosys, Pharmacyclics, Roche/Genentech, Symbio and TG Therapeutics; speakers fees from Astellas; and research funding to his institution from AbbVie, AstraZeneca, Celgene, Epizyme, Gilead, TG Therapeutics and Trillium.