Introduction

Hodgkin lymphoma (HL) is a hematological malignancy involving the lymphatic system. The crude incidence of HL in the European Union is 2.3/100,000/year [1]. Most patients are diagnosed between 15 and 30 years of age, followed by another peak in adults aged ≥ 55 years [2, 3]. About 20% of the patients are refractory or relapse after chemotherapy, radiation therapy or their combination. The overall survival for patients with refractory HL remains poor; therefore, the highest chances of survival for those patients are clinical trials and the development of novel therapeutic modalities that could yield durable remissions and improved survival [4]. Based on its histological characteristics (a rather small number of primary tumor-associated CD-30+ Reed Sternberg cells surrounded by a granuloma-like, immune cell-rich environment) HL is an ideal candidate for anti-PD-1 therapy [5]. Therefore, it is not surprising that monoclonal antibodies that block the interaction between PD-1 and PD-L1, by binding to either the ligand or receptor, have shown remarkable activity in HL [6,7,8,9,10,11,12,13,14,15]. A growing number of clinical trials evaluating different immune checkpoint inhibitors in classical HL in various treatment settings and in various combinations with other agents is ongoing [16, 17].

The aim of this paper was to provide an overview, selecting the most relevant papers on the role of imaging in patients with HL treated with immune checkpoint inhibitors.

The present paper is descriptive, with the objective of providing current status and future directions for the investigations in the field; therefore, the recommendations for a systematic review cannot be applied [18].

Checkpoint inhibitors

Programmed-death 1 receptor (PD-1, CD279) is one of the crucial molecules that turn down the activation of the immune response functioning as negative regulator of T cell [19, 20]. The PD-1 receptor binds two ligands the PD-L1 (CD274, B7-H1) and the PD-L2 (CD273, B7-DC) [19]. PD-1 expression typically increases as cells are activated and signaling through PD-1 by its ligands results in cells becoming senescent with an exhausted phenotype. A subset of cells subsequently becomes apoptotic [21]. In the tumor microenvironment, PD-1 expression inhibits numerous immune cell subsets including T cells, B cells, natural killer cells, dendritic cells, and tumor-associated macrophages [22]. The activation of the PD-1/PD-L1 immune checkpoint pathway in cancer represents an adaptive mechanism of resistance used by cancer cells against tumor-infiltrating lymphocytes [5]. Immune checkpoint ligands, PD-L1 and PD-L2, are expressed on the surface of malignant cells in 65–100% of classical HL [7, 23, 24] and in 54% of nodular lymphocyte-predominant HL [24]. Alterations of the PD-L1/PD-L2 locus (present in 97% of cases) have been found to define classical HL and to be associated with patient outcome in a study on 108 newly diagnosed classical HL [23]. The presence of PD-1 ligands, predominantly on the Reed-Sternberg cells, and also on immune cells within the tumor microenvironment, as well as the expression of PD-1 receptors on intratumoral T cells, suggests significant suppression of T-cell function due to PD-1/PD-L1/PD-L2 interactions, providing the rationale for the use of PD-1 blockade to reverse the T-cell inhibition and to allow for a more effective antitumor immune response in HL [21, 23]. Figure 1 summarizes the mechanisms of action of the immuno-therapeutic agents.

Fig. 1
figure 1

Mechanisms of action of PD-1 blocking agents

Full size image

Both nivolumab and pembrolizumab are fully humanized IgG4 anti-PD-1 monoclonal antibodies. Nivolumab (Opdivo®, Bristol-Myers Squibb) is approved in both Europe and United States (US) as monotherapy for the treatment of adult patients with relapsed or refractory classical HL after autologous stem cell transplant (ASCT) and treatment with brentuximab vedotin. Pembrolizumab (Keytruda®, Merck & Co.) is used in Europe in adult classical HL as monotherapy after brentuximab vedotin and ASCT failure, or when after brentuximab vedotin failure transplant is not possible. In US, it is approved for the treatment of adult and pediatric patients with refractory classical HL, or who have relapsed after three or more prior lines of therapy.

Treatment response criteria in HL

The assessment of response to initial treatment in HL is crucial because the need for additional treatment is based on the treatment response. Many guidelines have been proposed to evaluate treatment response criteria in HL (Fig. 2).

Fig. 2
figure 2

Timeline depicting the use of treatment agents/modalities and the development of response assessment guidelines in lymphoma

Full size image

The Cotswold classification, dating back to 1989, first formally defined the complete remission unconfirmed (CRu) to describe cases presenting with a residual mass after treatment that was most probably fibrotic [25].

In 1999, a multidisciplinary panel of experts constituted the International Working Group (IWG) and published guidelines for non-Hodgkin lymphoma (NHL) response assessment and outcomes measurement [26]. The IWG criteria (i.e., Cheson 1999 criteria) included anatomic definitions of response, with normal lymph node size after treatment of 1.5 cm in the longest transverse diameter on computer tomography (CT). A designation of complete response/unconfirmed was adopted to define patients with a greater than 75% reduction in tumor size after therapy but with a residual mass [26]. These recommendations rapidly entered in clinical practice and were used in the approval process for a number of new agents. However, they were subject to misinterpretation and recommended technologies (e.g., gallium scan), which were not the state-of-the-art modalities for lymphoma patients assessment. In 2007, the IWG guidelines were revised. This updated version (i.e., Cheson 2007) included “new” technologies such as immunohistochemistry and 18F-fluorodeoxyglucose–positron emission tomography/CT (FDG PET/CT) to define treatment response [27]. Although these criteria were developed for the end-of-treatment evaluation, they have been used for the interim response assessment as well [2]. However, the 2007 criteria were still subject to misinterpretation, since there was significant potential for ambiguity in the evaluation of the PET scans, which was based on a subjective interpretation of what represented [18F]FDG “background” activity (e.g., blood pool versus adjacent regions) and the degree of significantly discernible uptake, compared to the background [28]. In 2009, the Deauville criteria were introduced to interpret both interim and end-of-treatment FDG PET/CT in HL. These criteria adopted a visual 5-point scale (5-PS) to determine the degree of residual FDG lesion uptake in comparison to that of mediastinum and liver (i.e., reference tissues) [29]. In 2014 an updated version of the Deauville classification (i.e., Lugano 2014 criteria) was published [30]. The goal of the Lugano criteria was to reduce ambiguity and to achieve more consistent therapeutic response assessments for patients enrolled in clinical trials [28]. They were the direct consequence of the need of integration of the Deauville criteria and the input from the investigators at the International Workshop Conferences in 2011 and 2013 [28, 31]. According to the Lugano criteria patients could be classified as complete (Deauville 1–3) or not complete (Deauville 4–5) metabolic responders. The Lugano criteria were based on experience with chemotherapy or chemoimmunotherapy; however, the clinical availability of an increasing number of biologic agents with immune mechanisms required new criteria in image interpretation to account for these agents’ biologic or immunomodulatory effects. Therefore, in 2016, the Lugano criteria were refined (i.e., Lymphoma Response to Immunomodulatory Therapy Criteria–LYRIC 2016 criteria) introducing the concept of “indeterminate response” for those cases with “delayed response” or “pseudo-progression” that may be related to recruitment of immune cells to disease site [32]. In the case of indeterminate response, biopsy or repeat imaging is suggested to re‐classify the disease as either true or pseudo-progression.
In the same year, aiming at the harmonization of the lymphoma response criteria with response evaluation criteria in solid tumors (RECIST) were proposed and approved the Response Evaluation Criteria in Lymphoma (RECIL 2017). Also the RECIL guidelines include the “pseudo-progression” definition in case of a treatment with immunomodulating agents, new immunotherapies, and cell therapy with chimeric antigen receptor engineered T cells [33].

The image-based management of HL during treatment in clinical practice: the ESMO and the NCCN Guidelines

According to the European Society of Medical Oncology (ESMO) Guidelines neck/chest/abdomen contrast-enhanced CT (CECT) and FDG PET/CT should be performed at staging. The interim response evaluation by CECT should be carried out after completion of chemotherapy/before RT in limited and intermediate stages, while in advanced stages it should be performed after four cycles of chemotherapy and before radiation treatment (RT). Interim FDG PET/CT may be useful to identify poor-risk individuals; however, it cannot be considered standard and should be restricted to clinical trials except for the decision of whether patients with advanced HL receiving escalated combination of bleomicin, etoposide, adriamycin, ciclophosphamide, oncovin, procarbazine, and prednisone (BEACOPP) require RT. Final assessment should be carried out after completion of treatment. While CECT is mandatory, FDG PET/CT should be carried out whenever this diagnostic tool is available [34].

According to the National Comprehensive Cancer Network (NCCN) Guidelines neck/chest/abdomen/pelvis CECT scan should be performed at baseline and 6, 12, and 24 months after treatment completion, or as clinically indicated. FDG PET/CT should be performed at baseline. The value of interim FDG PET/CT remains unclear for many clinical scenarios; therefore, all definition of interim-response should be considered in the context of management decisions. The guidelines recommend biopsy for all patients with Deauville score 5 and all patients with a positive biopsy should be managed as described for refractory disease. For those with a negative biopsy, complete response (CR) should be documented, including reversion of PET to “negative” within 3 months after therapy completion. Surveillance FDG PET/CT should not be done routinely due to risk of false positives. Management decisions should not be based on FDG PET/CT scan alone; clinical or pathologic correlation is needed. FDG PET/CT should be performed during the follow-up only if the last scan was scored as a Deauville 4–5. The NCCN Guidelines recommend that the Deauville score, essential to decide how to manage the patient, should be included in all nuclear medicine reports [2].

Neither the ESMO nor the NCCN Guidelines mention that criteria for both CECT and FDG PET/CT treatment response evaluation may not be adequate for anti-PD-1 treatment response assessment.

Treatment response to anti-PD-1 in HL: the role of imaging

Assessing response to treatment in HL is classically based on bi-dimensional measurement of lymph nodes and masses although, as mentioned above, in the past 10 years guidelines have been updated to reflect the value of FDG PET/CT imaging in defining metabolic remission as equivalent to anatomical remission [35]. The added value of FDG PET/CT over conventional morphological imaging is the ability to identify viable tumor in residual lesion(s) [36]. However, the rate of false positive results of FDG PET/CT after treatment is not negligible. Nonetheless, experience in solid tumors has led to the recognition that immunomodulatory therapies, including the immune checkpoint inhibitors, may lead to tumor behaviors not well captured by standard imaging criteria [35, 37]. Specifically, the concern is that some patients receiving immune checkpoint inhibitors demonstrate “tumor flares” that do not have the same prognostic implications as progressive disease (PD). Figures 3 and 4 show examples of two patients with disease classified as indeterminate response. This new “indeterminate response” category, introduced to ensure that patients do not discontinue immune checkpoint inhibitors before receiving maximal benefit, includes patients in whom the tumor grows within the first 12 weeks of therapy without clinical deterioration; a single lesion grows at any time during therapy but overall tumor burden does not increase, or PET/CT imaging shows an increase in FDG uptake without a change in tumor size. Repeat imaging is suggested and, if growth continues, the patient is considered to have PD [32]. However, it is not yet established whether the assessment of “indeterminate response” translates in a different survival outcome than the traditional assessment of PD [35, 38] and the efficacy of the immune checkpoint inhibitors was demonstrated using the Cheson 2007 criteria [27] in almost all the prospective clinical trials (Table 1). More recently, the Lugano criteria have been applied [11, 39, 40].

Fig. 3
figure 3

FDG PET/CT scans (a, e and i—PET maximum intensity projection; b, f and j: axial PET; c, g and k: axial CT; d, h and l—axial fused PET/CT) performed at baseline (a, b, c and d), 3 months (e, f, g and h) and 5 months (i, j, k and l) from anti-PD-1 (nivolumab) treatment initiation, of a 30-year-old female, affected by classical Hodgkin Lymphoma, stage IIA at diagnosis, refractory to chemotherapy with persistence of disease at mediastinum and left latero-cervical node (green arrows), as shown on baseline FDG PET/CT scan. At 3 months’ evaluation the baseline lesions were no more detectable, while a carinal lymph node (red arrows) with FDG uptake (Deauville score 4) was visible and classified as “indeterminate response”. At 5 months’ evaluation the carinal lymph node was confirmed, with increased FDG uptake (Deauville score 5). *FDG uptake related to an artifact

Full size image
Fig. 4
figure 4

FDG PET/CT scans (a, h and o—PET maximum intensity projection; b, e, i, l, p and s: axial PET; c, f, j, m, q and t: axial CT; d, g, k, n, r and u: axial fused PET/CT) performed at baseline (a, b, c, d, e, f and g), 4 months (h, i, j, k, l, m and n) and 6 months (o, p, q, r, s, t and u) from anti PD-1 (nivolumab) treatment initiation, of a 37-year-old male, affected by classical Hodgkin Lymphoma, stage IIA at diagnosis, refractory to chemotherapy with persistence of disease at nodal supra and sub diaphragmatic level (green arrows), as shown on baseline FDG PET/CT scan. At 4 months’ evaluation the baseline lesions were no more detectable, while a pulmonary lesion (red arrows) with FDG uptake (Deauville score 4) was visible and classified as “indeterminate response”. At 6 months’ evaluation the pulmonary finding was no longer present on both PET and CT images

Full size image
Table 1 Summary of prospective clinical trials’ results that proved the efficacy of the immune checkpoint inhibitors in Hodgkin Lymphoma
Full size table

In the CheckMate-039 clinical trial 23 patients with relapsed/refractory HL were treated with nivolumab. Both the best overall response (i.e., the best response between the date of the first dose and the last efficacy assessment before subsequent therapy) and the objective response rate (i.e., the proportion of the total number of patients whose best overall response was either a partial (PR) or a complete response) were assessed by CT scan (FDG PET/CT was used to confirm a CR). The objective response rate was 87% (4 CR and 16 PR) [6]. In the Checkmate-205 clinical trial recently updated, 243 patients were treated with nivolumab. The primary outcome of the study was to evaluate the proportion of patients achieving an objective response, defined as the percentage of treated patients with a best overall response of complete or partial remission (assessed by an independent radiological review committee). Patients were assessed for tumor responses by CT (preferred) or MRI and FDG PET/CT. For patients with bone marrow involvement at screening, a bone marrow biopsy was required to confirm complete remission [7]. After a median follow-up of 18 months, 40% of patients continued to receive treatment. An objective response was observed in 168 patients (40 CR and 128 PR) with a median duration of response of 16.6 months, and a median progression-free survival of 14.7 months [15]. A higher percentage of objective response (4 CR and 9 PR with a median time to response of 8 weeks) was observed in 17 patients treated within the Japanese nivolumab phase II study. The objective response ratio (primary study endpoint) was assessed by a central review committee using CT (preferred) or MRI and FDG PET/CT [8]. In the KEYNOTE-013 study were enrolled 31 HL patients with the aim of assessing the CR rate at any time after treatment with pembrolizumab. An objective response was observed in 20 patients (5 CR and 15 PR) [9]. The KEYNOTE-087 study enrolled and treated 210 patients to evaluate the overall response rate to pembrolizumab in three different cohorts of patients with refractory/relapsed HL (i.e., disease progression after autologous stem cell transplant (ASCT) with/without brentuximab vedotin and salvage chemotherapy plus brentuximab vedotin). Response was assessed by CT scan by an independent radiological review committee. FDG PET/CT was performed to confirm CR or PD and as clinically indicated. The overall response rate was higher than that observed in the KEYNOTE-013 study with 145/210 responders (47 CR and 98 PR) [10, 41].

The efficacy of low-dose pembrolizumab was evaluated in 5 HL patients with excellent results. Treatment response was evaluated by FDG PET/CT applying the 5-PS according to the Lugano 2014 criteria. All treated patients had an objective response (4 CR and 1 PR) [11].

Anti-PD1/PD-L1 in HL: the search for valuable biomarkers

Since the 2000s, the assessment of lymphoma has been essentially based on clinical examination, imaging, and bone marrow biopsy [36]. The advent of immunotherapies has opened different scenarios and has given rise to new needs. With innovative expensive targeted therapies for lymphomas, the need for accurate staging systems and standardized criteria for response is even more critical [36]. Therefore, biomarkers to selectively identify best candidates are necessary [42, 43].

Blood and tissue biomarkers

It is clear from the clinical results obtained to date that PD-1 blockade by itself is not effective in all patients with responsive tumors and even in those with response, delayed or mixed tumor regression can be seen. This is related to the dynamic nature of the immune system and the number of elements involved in the complex immune response against cancer [44]. The high response rate to anti-PD-1/PD-L1 therapy in HL could not be directly dependent on high ligand expression but on the effect that this high ligand expression may have on the microenvironment [22, 43]. Therefore, developing biomarkers for immuno-therapeutics is more challenging than developing biomarkers for targeted-therapy [45]. Immunohistochemical detection of PD-L1 on tumor cells is so far the most common biomarker for patient selection and prediction of response to anti-PD-1/PD-L1 therapy in clinical practice [46]. In fact, PD-L1 overexpression is expected to predict a better response to checkpoint inhibition and improve patient selection [45]. However, there are some challenges associated with the cell detection of PD-L1. First, there is no a standard definition of the cut-off to be considered overexpression and the use of different assays prevents the direct comparison of the results [45, 47]. Additionally, the PD-L1 expression on tumor and immune cells is a dynamic process that can also be regulated by intrinsic oncogenic pathways, and may not be reflected by a single time point evaluation [20, 45, 48].

The identification of the genetic signatures (e.g., 9p24.1 gene translocations or amplifications) by DNA sequencing techniques has been proposed to identify patients who have higher chances to respond to checkpoint inhibition [45, 49]. As mentioned above, the tumor microenvironment could predict response to checkpoint inhibitors, and tumor-infiltrating immune cells may be examined by immunohistochemistry or flow cytometry [43, 45].

Profile of the circulating cell-free DNA has been described in a proof-of-concept study to vary rapidly at treatment initiation suggesting a potential role for response monitoring [50]. Recently, longitudinal ctDNA profiling has been demonstrated to identify treatment-dependent patterns of clonal evolution in 80 newly diagnosed and 32 refractory HL patients relapsing after chemotherapy and in those maintained in PR under immunotherapy. Moreover, the authors affirmed that quantification of ctDNA complemented interim PET/CT in response assessment [51].

In some cancer types, virtually each tumor is virus-associated, while in other cancers including HL, only a subset of cases is virus-associated. Therefore, the presence or absence of the causative virus (i.e., Epstein-Barr virus in HL) could represent a predictive biomarker for response to immune checkpoint blockade. Alternatively, measures of endogenous immune responses to these viruses in patients bearing virus-positive tumors could serve as a predictive biomarker [49].

Imaging biomarkers

The need to find reliable biomarkers is also based on the need for sustainable healthcare costs since an antibody-based treatment is expensive (e.g., a single course of nivolumab costs between $100,000 and $150,000) [52]. Therefore, biomarkers derived from conventional imaging, antibody-based imaging or other imaging-based approaches could be valuable to select patients suitable for checkpoint inhibitors treatment. Recently, Dercle et al. [53] showed that 3 month-FDG PET/CT was able to detect all HL patients responding to immune-checkpoint blockade by anti-PD1 treatment (responders experienced a significant shrinkage in tumor volume—∆MTV, ∆TLG—, a decrease in tumor glucose metabolism and an increase in spleen metabolism) [53]. These results appear to be in line with the immunohistochemical finding of the association between glucose transporter 1 (GLUT1) and PD-L1/PD-L2 expression [54]. Nonetheless, even though the role of functional imaging parameters appears promising it should be validated in larger series. Figures 5 and 6 show two cases of patients with CR and PR, respectively, on interim FDG PET/CT. However, as mentioned above, tumor behaviors related to immunomodulatory therapies may not be well-captured by standard criteria applied to conventional imaging. The main issue of applying standard imaging criteria to immune checkpoint molecules treatment assessment is related to the fact that the mechanism of these new drugs is a complex and dynamic process, only partly understood and different among tumor types.

Fig. 5
figure 5

FDG PET/CT scans (a and e: PET maximum intensity projection; b and f: axial PET; c and g: axial CT; d and h: axial fused PET/CT) performed at baseline (a, b, c and d), and 4 months (e, f, g and h) from anti PD-1 (nivolumab) treatment initiation, of a 45 year-old male, affected by classical Hodgkin Lymphoma, stage IIIXB at diagnosis, refractory to chemotherapy with persistence of disease at node supra and sub diaphragmatic level (green arrows), as shown on baseline FDG PET/CT scan. At 4 months evaluation the baseline lesions were no more detectable (Deauville score 1). Therefore, the patient was classified in “complete response”

Full size image
Fig. 6
figure 6

FDG PET/CT scans (a and e: PET maximum intensity projection; b and f: axial PET; c and g: axial CT; d and h: axial fused PET/CT) performed at baseline (a, b, c and d), and 6 months (e, f, g and h) from anti PD-1 (nivolumab) treatment initiation, of a 28 year-old male, affected by classical Hodgkin Lymphoma, stage IVB at diagnosis, refractory to chemotherapy with persistence of disease at node supra and sub diaphragmatic level and at the iliac left bone (green arrows), as shown on baseline FDG PET/CT scan. At 6 months’ evaluation the baseline most of lesions were no more detectable (Deauville score 1) while some inter-cavo-aortic lymph nodes displaying FDG uptake were still present (Deauville 4). Therefore, the patient was classified in “partial response”

Full size image

In this regard, PET/MR modality adding functional information derived from MR technologies, especially diffusion weighted imaging (DWI), can provide additional information about tumor cellularity improving the characterization of the tumor phenotype [36].

Molecular imaging has been at the forefront of non-invasive assessment “in vivo” of the tumor phenotype during the past decade and allows to image tumors using designable imaging agents [52]. The improvements in radiochemistry and isotope development, nanomedicine and nanotechnologies, made possible to design PET imaging agents suited for a specific target. Therefore, the possibility to image patients by immunoPET with PD1/PD-L1-radiopharmaceuticals to visualize “in vivo” the PD-L1 expression and its variation over time might be realistic in the next future.

Recently, a renewed interest has been shown in radiomics. It is based on the concept that biomedical images contain more information than that provided by qualitative analysis and that this information, reflecting the underlying pathophysiology, can be revealed via quantitative image analyses [55]. In fact, up to hundreds of image-derived parameters can be calculated from CT, PET/CT and MR, to be tested for correlations with biological features. As described above, current conventional imaging practice is generally qualitative or, when quantitative, measurements are commonly limited to one or bi-dimensional assessment of tumor size. However, these measures do not reflect the complexity of tumor heterogeneity or behavior, nor, in many cases, are changes in these measures predictive of therapeutic benefit [55], especially in the immunotherapies era. The modern re-birth of radiomics aims to convert images into mineable data, with high fidelity and high throughput [55]. Literature data on radiomics in HL focused on chemotherapy response prediction using either CECT [56] or PET/CT imaging [57] are promising, but still not definitive. However, initial experiences in solid tumors suggest that radiomics may predict early failure to nivolumab and may help in the selection of patients who may benefit from PD-1/PD-L1 treatment [58, 59].

Conclusions

The emergence of modern more individualized therapeutic approaches in HL needs to be accompanied by robust predictive models and reliable response assessment biomarkers and criteria. Immune checkpoint inhibitors represent a major advance in the treatment of relapsed/refractory HL patients. Initial attempts to face the uncertainties related to the assessment of treatment response have been made, but further investigations for the evaluation of the role of both immuno-histochemical and imaging biomarkers are needed. Novel blood, tissue, and image-derived parameters have shown promising potential. These, even in combination with conventional risk factors, should be validated. To provide robust data for response definition, concomitantly with new drug approval, a multi-biomarker monitoring should be incorporated into clinical trials. This strategy should allow to precisely and timely define biomarkers cumulative accuracy in assessing the disease status of HL patients, and at long term, to improve outcome and optimize resources use.