In this issue of the EJNMMI, Seban et al. [1] propose a prognostic score combining baseline total metabolic tumor volume (TMTV) and the Derived Neutrophil-to-Lymphocyte Ratio (dLNR) to predict overall survival (OS) in non-small cell lung cancer (NSCLC) patients receiving immune-checkpoint inhibitors (ICI). In their series of 80 patients, a TMTV >75 and a dLNR >3, previously validated as a biomarker of immune activation by other groups [2], resulted as independent predictive factors for shorter OS and worse clinical benefit, the later one established according to RECIST 1.1 criteria and combining objective response (partial or complete) at any time during the course of ICI or stable disease after 6 months of treatment. The same group from Gustave Roussy Institute, had shown in melanoma patients that baseline 18F-FDG PET parameters could be used to build a prognostic metabolic score using not only the TMTV but also hematopoietic tissue metabolism, namely, bone marrow to liver ratio (BLR) proving that these PET metrics had significant and independent value in predicting survival [3]. Hence, high-risk patients (high TMTV and high BLR) had worse survival compared to low-risk patients (low TMTV and low BLR). In this cohort, BLR not only outperformed spleen to liver ratio (SLR), which has been showed by others groups to predict OS in melanoma [4] but was additionally associated with transcriptomics profiles, including regulatory T cells markers.

Along with the prognostic role of the above-mentioned baseline parameters, recent data reported by another group [5] suggest a higher risk of hyper progressive disease during immunotherapy in NSCLC patients presenting with high metabolic tumor burden, expressed by both MTV and TLG, and elevated inflammatory indexes, including dNLR and platelet count. As expected, multivariate analysis identified once again MTV and dNLR as independent predictors for OS.

Of note, although tumor SUVmax was not found as a significant predictor of survival by Seban et al. [1], it resulted correlated to the PD-L1 expression status. This finding is in line with what previously suggested by other groups [6, 7], the hypothesis being that increased glycolytic activity can be linked to an enhanced immune infiltrate in the stromal compartment. Paradoxically, NSCLC patients presenting with a higher baseline SUVmax could have a higher chance of response in case of treatment with ICI. This initially theory [6] has indeed anticipated the results published only recently by Takada et al. [8] in 89 patients with advanced or recurrent NSCLC reporting a significantly higher response rate in patients with a baseline SUVmax ≥ 11.16 (41.3%) compared to patients with lower SUVmax values (11.6%).

When it comes to 18F-FDG PET/CT for therapy monitoring, the nuclear medicine community has been very active in the last decade, coming up with various response criteria [6, 7] including the PERCIMT (PET Response Evaluation Criteria for Immunotherapy) [9], PECRIT (PET/CT Criteria for Early Prediction of Response to Immune Checkpoint Inhibitor Therapy) [10], iPERCIST [11], or other PERCIST (PET Response Criteria in Solid Tumors)-adapted criteria [12]. A good example is represented by imPERCIST5 [13], developed by the Memorial Sloan-Kettering Cancer Center group in patients receiving ipilimumab. As opposed to PERCIST, where a new metastatic lesion classifies the patient as progressive metabolic disease (PMD), imPERCIST5 incorporates a new lesion in the 5 target lesions and requires the variation in the sum of SUVpeak among these 5 target lesions to be greater than 30% to define disease progression. While one might argue that the rate of pseudoprogression is lower with anti PD1/PDL1 mAbs than with Ipilimumab, an anti CTLA4, these criteria could be useful in patients receiving a combination of anti CTLA4 and anti PD1/PDL1 [14], or could be tested in those patients with a progression as per PERCIST on early interim PET while achieving a clinical benefit. Indeed, this situation has been recently reported to occur in more than one third of the NSCLC patients treated with Nivolumab or Pembrolizumab and classified as PMD on early interim PET [15]. Description of the clinical benefit experienced by patients with atypical evolutive patterns described in the paper from Humbert and colleagues [15], along with the concept of treating patients beyond objective progression [16, 17] should probably encourage our community to a certain humility when interpreting 18F-FDG PET scans until criteria are properly validated and accepted among the many PERCIST-adapted criteria proposed so far (Fig. 1a). To convince the medical oncology community, it is time to pool data and come up with user-friendly and widely adapted criteria.

Fig. 1
figure 1

Representative examples of PET metrics, response monitoring criteria (a) and important patterns to be sought for on follow-up scans (b) in patients receiving immune checkpoint inhibitors (ICI). MATV: metabolic active tumor volume; dLNR: derived lymphocytes to neutrophils ratio; PDL1: programed death ligand 1; irAEs: immune related adverse effects, PERCIST: PET Response Criteria in Solid Tumors; PECRIT: PET/CT Criteria for Early Prediction of Response to Immune Checkpoint Inhibitor Therapy; PERCIMT: PET Response Evaluation Criteria for Immunotherapy; ICI: immune checkpoint inhibitors

Full size image

Based on the recent literature, it is also obvious that the PET community should seek more than “just” tumor response [18], using baseline parameters, identification of signs of immune activation and immune-related adverse effects (irAEs) on follow-up scans (Fig. 1b). As nicely shown by Seban and colleagues, combined criteria or prognostic scores are useful when PET metrics and combined biological parameters are demonstrated to be independent prognosticators.