The value of FDG PET/CT imaging in outcome prediction and response assessment of lymphoma patients treated with immunotherapy: a meta-analysis and systematic review

Treatment strategies of the lymphoid malignancies were revolutionized by immunotherapy in two major advances over the last 3 decades. In 1997, the introduction of anti-CD20 monoclonal antibodies (mAbs) such as rituximab which targeted B-cells exclusively to evoke a direct anti-tumor cytotoxic effect revolutionalized the treatment paradigm for lymphoma. Then, in 2017, the introduction of immune checkpoint inhibitors (ICI) such as anti-PD1 and anti-CTLA4 monoclonal antibodies which stimulates the immune system via T cells appears to be the next promising step in lymphoma management [1‐3]. Although immunotherapy now is common terminology for the class of drugs that stimulate the immune system to indirectly target cancer cells, conventionally, immunotherapy refers to any therapy that manipulates the immune system such as both anti-CD20 mAbs and new targeted immunotherapy, ICIs, and cell therapy.

Anti-CD20 monoclonal antibodies have FDA approval for treatment of non-Hodgkin lymphoma (NHL) for multiple indications as monotherapy or in combination with other lymphoma-directed therapeutics [4, 5], and have a place in treatment paradigms for many B cell lymphomas [6]. Among the immune checkpoint inhibitors, pembrolizumab and nivolumab have regulatory approval for the treatment management of the relapsed or refractory transplant ineligible or post-transplant relapsed classical Hodgkin lymphoma (HL) and relapsed or refractory primary mediastinal B cell lymphoma (PMBL) [7, 8]. CD19-directed chimeric antigen receptor (CAR) T cell therapy is approved for the treatment of certain types of large B-cell lymphoma relapsed or refractory to at least two other treatment regimens [9].

Because of the inherent property of HL and most subtypes of NHL as highly ¹⁸F-FDG avid tumors, functional ¹⁸F-FDG PET/CT imaging is a standard response assessment tool for these diseases and has a decisive role in noninvasive response monitoring of therapies, including immunotherapeutics [10, 11].

In the abovementioned framework, multiple different visual ¹⁸F-FDG PET criteria were introduced to more consistently evaluate ¹⁸F-FDG PET scans [10, 12]. Despite the available ¹⁸F-FDG PET criteria, following the introduction of ICIs some modified response assessment criteria were proposed. These criteria were developed based on the observation that some immunomodulatory drugs can alter tumoral glucose metabolism, changing the assumed association between the ¹⁸F-FDG uptake and treatment efficacy observed under conventional chemotherapy [2].

As evident from the literature, existing original articles have some differences in the population and methodologies. Specifically, the malignant lymphoma analysed (either HL or NHL), immunotherapy regimens and line of therapy (anti-CD20 mAbs, ICI versus immune cell therapy), and time points and time intervals of performing ¹⁸F-FDG PET scan (baseline, early, and late ¹⁸F-FDG PET imaging), all vary considerably.

Various studies have dealt with the issue of the value of metabolic imaging by ¹⁸F-FDG PET scan for immunotherapy management in malignant lymphoma patients, but the evidence for lymphoma treated with ICIs is lacking due to the current standard being that of a quantitative or semi-quantitative assessment, primarily using the Lugano or LYRIC criteria [13, 14]. Here, we will review the role of ¹⁸F-FDG PET imaging in response assessment or response prediction in the lymphoid tumors treated with immunotherapy regimens using the perspective of the mentioned studies.

The performed search approach is presented as a PRISMA flowchart [107] in Fig. 1. From the initial 1488 identified papers, 91 were ultimately included in our study. As we expected, the eligible included studies showed discrepancies in methodological aspects and the main remarkable one was the applied treatment strategy. In the first step, the enrolled studies were classified into studies on anti-CD20 monoclonal antibodies (for example, rituximab), studies on ICIs (for example, nivolumab), and studies on cellular therapies (for example, CAR T-cell therapy or dendritic cell therapy). The former part involved 77 publications, whereas the two latter ones included a total of 14 investigations. Basic study characteristics of each part are separately listed in three tables (Supplemental Table 234) [16‐106].

The high sensitivity of the ¹⁸F-FDG PET in detecting nodal and extranodal involvement has established its role in primary staging as a standard of care for all ¹⁸F-FDG-avid lymphomas [11]. Calculating quantitative baseline tumor burden and tumor metabolism indices following immunotherapy are common parameters used to predict prognosis in baseline ¹⁸F-FDG PET scans [25, 27, 32, 104, 105]. However, some studies have not found a definite correlation between one or more of these parameters with PFS or OS [37, 95]. This can, in part, be explained by heterogeneous patient characteristics, including a wide age range, different clinical stage and disease subtypes, and different software and the different ways used for definition of the marginal threshold of hypermetabolic foci [27]. Moreover, these discrepancies are obvious through the heterogeneity of reported HRs in Figs. 2 and 3. Based on the meta-analysis performed in the present work, the highest pooled HR between baseline parameters belongs to MTV with HR of 4.39 (95%CI: 2.71–7.08; P = 0.000] and the lowest one belongs to SUV_max with an HR of 1.18 (95%CI: 0.79–1.75; P = 0.404).

Many reports suggest response assessment via semi-quantitative/quantitative methods in lymphoma patients provides better outcome discriminators than the visual criteria regardless of the considered background reference tissue [21, 22, 38, 48]. Moreover, it has been reported that inter-observer reproducibility was higher using semi quantitative methods than the visual approaches [48]. Our findings showed the pooled HR of the early response assessment following anti-CD20 treatment using ΔSUVmax for PFS is 3.25 (95%CI: 2.08–5.08; P = 0.000), which is higher than the DS and IHP criteria’s pooled HRs. Pooled HR for an early evaluation concerning OS did not support this point and the pooled HR of the ΔSUVmax is lower than the DS corresponding value, 2.87 (95%CI: 1.92–4.28; P = 0.000] versus 3.23 (95%CI: 1.87–5.58, P = 0.000). On the other hand, according to some studies, it seems that visual assessment by DS is a better prognosticator than the IHP criteria [45]. Our findings in early response assessment pooled HRs also support this point. Pooled HR of the early response assessment for PFS and OS are respectively 2.56 (95%CI: 1.79–3.66; P = 0.000] and 3.23 (95%CI: 1.87–5.58; P = 0.000) for DS, which are higher than the IHP criteria’s pooled HRs. These measurements for DS and IHP were relatively similar for EOT ¹⁸F-FDG PET/CT. This difference can in part be explained by the more conservative nature of the IHP criteria, which uses a lower visual reference (surrounding background or mediastinal blood pool) compared to the DS, which uses liver parenchymal uptake [18]. It should be mentioned that the SUV of the reference organs may be affected by the hypermetabolic tumor burden, and this point should be considered when interpreting serial ¹⁸F-FDG PET/CT scans [51].

In immunotherapy with ICIs, a decrease in tumor metabolism indices as early as 8 weeks after therapy initiation occurred in responders. On the other hand, modifications of tumor burden indices occurred appreciably later. This time interval may be due to immune system reactivation and glucose consumption by tumor-infiltrated lymphocytes [24]. It seems that immune-related adverse events in immune cell therapy and ICIs have more impressive influence on 18F-FDG PET/CT scan, compared to the anti-CD20 immunotherapy. There are more anti-CD20 monoclonal antibody studies performed which allowed the analysis, while the current published studies in ICIs and cell therapy are quite heterogeneous with regard to response assessment and outcome prediction; therefore, dedicated analysis in this treatment cohort was not performed.

The number of enrolled papers on anti-CD20 monoclonal antibodies is quite different, compared to the other two categories: ICIs and cellular therapies. The latter ones have a small number of papers and do not have similar indices in all of them. This difference gives different strength to the results related to the anti-CD20 mAbs. With the support of the large number of studies on anti-CD20 mAbs, we could calculate pooled HRs, whereas meta-analysis and pooled ratio calculation were not possible for ICIs and cellular therapies. On the other hand, in the category of anti-CD20 mAbs, the depicted funnel plots showed asymmetry and probable publication bias in pooled HR of some of baseline parameters (MTV, TLG, and SUVmax for PFS and MTV for OS) as well as pooled HR of interim response assessment using DS for OS. This was the main limitation that we encountered in the present study.

Our present study revealed that with anti-CD20 therapy, baseline MTV on ¹⁸F-FDG PET/CT has the highest HRs for both PFS and OS. In response assessment of anti-CD20 therapy, Deauville score in early and late time points has higher HRs for PFS and OS compared to the international harmonization project criteria. While early changes in ¹⁸F-FDG PET parameters may be predictive of response to ICIs and cell therapy treatment, further studies are required to establish the optimal response parameters following treatment of lymphoma patients.