In this issue of Blood, Mir et al1 demonstrated that 18F fluorodeoxyglucose (FDG) uptake level on pretreatment positron emission tomography/computed tomography (PET/CT) of de novo follicular lymphoma (FL) patients does not predict subsequent high-grade histological transformation (HT).
The ability to detect high-grade transformation of FL is important both for prognosis and for treatment decisions whereby patients may be offered anthracycline-based treatment. One of the difficulties is that when transformation occurs it is not uncommon to find both low grade and high grade in different areas of the body, and a biopsy from 1 site does not exclude the presence of high-grade disease elsewhere. HT is usually suspected based on clinical or imaging features. Examples of clinical features suggestive of HT include rapid growth of masses, rapid rise of lactate dehydrogenase, rapid developments of symptoms, and/or clinical deterioration or change in the tempo of previously indolent disease course. On the other hand, HT can also be suspected based on imaging features and frequently leads to significant discussions in tumor boards and multidisciplinary teams in everyday clinical practice.
The problem with suspecting HT on the basis of imaging is the lack of robust data to define diagnostic criteria and make sound judgments. Previous studies fall into 2 groups: those that compared uptake in low-grade vs aggressive lymphomas and those that compared uptake in transformed vs nontransformed low-grade lymphoma. In the first group, studies suggested that maximum standardized uptake value (SUVmax) cutoff of 10 to 13 can differentiate low-grade and aggressive lymphomas.2,3 Similarly, studies comparing cases with and without HT found a difference in SUVmax values.4-7 However, these studies were largely retrospective and small in size with a significant overlap of uptake ranges between groups.
The most noteworthy differences in the current study is the large sample size from the prospective phase 3 GALLIUM study and the design of the study, which examined the relationship between baseline SUVmax (bSUVmax) at time of diagnosis and subsequent HT later in the course of disease, with a median follow-up of 59 months. Previous studies compared uptake levels at the time of HT to uptake in nontransformed cases.
The authors found that neither the intensity of uptake (SUVmax) nor the variation of uptake level within same scan (SUVrange) predict subsequent HT in 549 patients with de novo FL with high tumor burden. Strikingly, there was a very wide range of SUVmax values in this large FL cohort with an upper value of 64.3 and in-patient variation in SUVmax between lesions of up to 59.8, although this could be partially explained by the inclusion of high-tumor-burden patients. Rates of subsequent HT were not significantly higher in patients with bSUVmax >10 or >20. Median SUVrange was not significantly different between patients with or without HT. These findings challenge the suspicion of HT in de novo cases based on uptake level alone. The main caveat of this study, which the authors acknowledge, is the lack of prescreening data for the GALLIUM study on the number of patients who had a biopsy, based on PET findings, which confirmed HT and were excluded from the study.
So what should we do when a new FL patient has a high SUVmax on PET/CT or there is significant variation of SUVmax values between different sites of the disease? This study shows that a high bSUVmax alone is not enough of an indication for a second biopsy. Perhaps less intuitive is the issue of having significant discrepancy of uptake between various sites in the same patient. Again, the study showed clearly that there is a very wide variability in FDG uptake between disease sites in FL, and this alone should not be an indication for additional biopsy from high-uptake sites.
There are 2 important points to highlight based on these findings. First, although the results of this study challenge the practice of repeating biopsy of high SUVmax areas, it is still advisable to direct the first biopsy to areas of highest SUVmax whenever feasible.
Second, it is worth emphasizing that these findings are driven from a population of untreated high-tumor-burden FL and do not necessarily apply to disease assessment at the time of relapse. What the study showed is that baseline uptake level did not predict subsequent HT, but this does not mean that at the time of developing HT the uptake level may increase significantly from its baseline; this could be a useful indication of the possibility of HT. At the time of relapse, comparison with previous imaging features is possible and adds more insight, and the risk of HT may be higher. A number of other imaging features may suggest HT (eg, change in tempo of disease, rapid growth in part of the disease, and/or significant increase in uptake intensity in some disease sites, development of necrosis, new development of multiple extranodal sites).
Accurately suspecting HT, to avoid both missing the diagnosis on one hand and unnecessary biopsies on the other hand, remains a difficult task. Future developments in imaging may help (eg, by applying the field of radiomics and machine learning to find imaging biomarkers that can predict HT). Until then, careful clinical judgment and examination of various features of the disease remain the mainstay of the decision to look for HT.
Conflict-of-interest disclosure: The author declares no competing financial interests.