Angiogenesis imaging study using interim [¹⁸F] RGD-K5 PET/CT in patients with lymphoma undergoing chemotherapy: preliminary evidence

Angiogenesis is a fundamental process involved in a variety of physiological and pathological conditions. The growth of solid tumours remains restricted to 2–3 mm in diameter until the onset of angiogenesis [1]. The concept of antiangiogenic therapy in clinical oncology is to stop cancer progression by suppressing the tumour’s blood supply, and more than 20 angiogenic growth factors have been studied, including their receptors and signal transduction pathways. This increasing use of targeted therapies has led to a growing demand for imaging the tumour response to these treatments. Such biomarkers not only would facilitate clinical trials of new drugs but could also be used to aid in the selection of optimal treatment for individual patients (personalised medicine).

Positron emission tomography–computed tomography (PET) using tracers for the assessment of glucose metabolism by [¹⁸F] fluorodeoxyglucose (FDG) is well established, and it plays a crucial role in initial staging and treatment response assessment, especially in patients with lymphoma [2]. Integrin αvβ3 is a heterodimeric transmembrane glycoprotein consisting of α- and β-subunits which plays an important role in angiogenesis [3]. These glycoproteins interact in cell–cell and cell–matrix interactions, easing endothelial growth and cell migration, therefore paving the way to angiogenesis [4]. Integrins assist in the progress of tumour development and metastasis by facilitating endothelial and tumour cell migration. An important phenomenon that depends on cell–extracellular matrix (ECM) interactions is the growth or sprouting of new blood vessels from a pre-existing vascular bed [4]. It has been found that several ECM proteins, like vitronectin, fibrinogen and fibronectin, interact with integrins via the amino acid sequence arginine–glycine–aspartic acid (RGD in the single letter code) [3]. Based on these findings, monomeric, multimeric and cyclic peptides, including the RGD sequence have been introduced to allow integrin αvβ3 imaging and picture pathological angiogenesis. Targeting specific angiogenesis molecular markers by PET imaging, like the integrin αvβ3, can be used for angiogenesis imaging. Li et al. recently showed that high RGD uptake on pre-treatment PET predicted antiangiogenic response in refractory patients presenting with solid cancer [5]. Patients presenting with solid tumours that showed low RGD uptake did not benefit from the antiangiogenic effect. Patients presenting with high RGD uptake benefited from the antiangiogenic effect. Therefore, RGD imaging could also be used for response assessment of antiangiogenic therapies.

[¹⁸F] fluorine arginine-glycine-aspartic (RGD-K5) PET/CT has been used in clinical studies on patients presenting with head and neck cancer [6], breast cancer [7] and lung cancer [8]. To the best of our knowledge, no clinical imaging study of angiogenesis in patients presenting with lymphoma has been performed.

The main objective of this study was to measure the impact of two cycles of standard chemotherapy on tumoural neoangiogenesis assessed by RGD-K5 PET/CT in patients presenting with lymphoma. The secondary objective was to analyse RGD uptake according to the lymphoma subtype.

The patients’ clinical data are summarised in Table 1. In all, 19 patients (7 women and 12 men) were included. Mean age at diagnosis was 55 ± 13 years. ECOG performance status was ≤ 1 for 18 patients (95%; ECOG performance status = 2 for the remaining patient). Nine patients had stage IV lymphoma (47%). The median follow-up time was 17.8 months (range 12–28). No patient died during or after treatment. One patient had synchronous follicular and classical Hodgkin lymphoma, thus both primary lesions were analysed. All 19 patients had both C0 FDG and RGD PET; 12 patients had both C2 FDG and RGD, completed the treatment protocol and were included for end-of-treatment analysis.

SUVmax for normal RGD organs is shown in Fig. 1 No statistical difference was found in normal organs before (C0) and after (C2) chemotherapy for SUVmax and SUVmean. High RGD uptake in the gallbladder (SUVmax 15.5 ± 6.7 (range 6.8–25.8) on C0 and 18.4 ± 9.6 (range 6.3–34.7) on C2, not shown in Fig. 1 for ease of visualisation), was due to radiotracer elimination.

Ex vivo analysis was performed on the initial biopsies of 10 patients. No correlation was found between endothelial cell marking via the ERGmarker, the number of vascular sections and C0 and C2 RGD uptake (data not shown).

RGD uptake was analysed according to histological subtype (Fig. 2). Apart from classical Hodgkin lymphoma (HL) and grey zone lymphoma (GZL), other lymphoma subtypes had low RGD uptake. RGD uptake was therefore compared between cHL and non-Hodgkin lymphoma (NHL), including grey zone lymphoma.

To illustrate these results, Fig. 3 shows information on a patient who had synchronous follicular and classical Hodgkin lymphoma. The patient had a history of cHL in 2014 which was considered in full remission after eight ABVD cycles, with early bone relapse in 2015. Bone relapse was successfully treated with four ICE cycles. The patient could not undergo autologous bone marrow transplant due to stem cell collection failure. In September 2017, the patient saw her haematologist for cervical swelling. Cervical biopsy revealed grade II follicular lymphoma. Initial FDG revealed lymph node and left sacral wing involvement. However, RGD uptake was only significant for the left sacral wing. The dissimilarity between FDG and RGD uptake was questioned with regard to the patient’s histopathological subtype. Bone biopsy was performed and confirmed concomitant bone cHL relapse.

Table 2 summarises FDG and RGD uptake at C0 and C2 for patients presenting with HL and NHL. At C0 and C2 FDG, there was no significant difference in SUVmax, SUVmean, MTV and ATV between patients with HL and NHL. At C0 RGD, patients with HL tended to have higher SUVmax, SUVmean and ATV than those with NHL. At C2 RGD, there was no significant difference in SUVmax, SUVmean and ATV between HL and NHL. An example is presented in Fig. 4.

Table 3 summarises FDG and RGD uptake at C0 and C2 according to final Deauville score. At C0 and C2 FDG and C0 RGD, there was no significant difference in SUVmax, SUVmean, MTV and ATV between responders and non-responders. At C2 RGD, non-responders tended to have higher SUVmax and SUVmean compared to responder (Fig. 5). There was no significant difference in RGD ATV between patients who responded to treatment or not.

To the best of our knowledge, this is the first clinical study to use RGD in patients with lymphoma. Patients presenting with cHL had significantly higher RGD uptake at baseline than NHL. Our prospective study shows that non-responders had higher RGD uptake at C2 RGD compared to responders.

RGD–K5 is a product that is expensive (around 1000 €) and difficult to synthesise. Due to technical problems in production, two patients were lost, two RGD PET scans were delayed and the investigator withdrew 10 potentially eligible patients. No patients had their treatment delayed because of RGD.

An end-of-treatment RGD PET would have allowed further description of angiogenic behaviour and changes and a more thorough analysis. Unfortunately, an end-of-treatment RGD PET was not performed for many reasons: logistics, cost, radiopharmaceutical difficulties and radiation exposure with no clear therapeutic implications.

As RGD is a new tracer, there is no clear reference or validation to determine the best threshold method to calculate ATV with RGD. However, Vera et al. showed that a fixed threshold of 1.4 SUV was the best for low tumour-to-background tracers [11]. As RGD has low uptake (Table 2), we tested this method to determine RGD ATV. However, the results were disappointing. We also tested all methods described in the paper published by Vera et al., with no success [11]. Thus, we decided to determine RGD ATV by visual sampling.

Our study allowed us to measure RGD uptake in normal organs before and after chemotherapy. Chemotherapy did not affect normal organ uptake [12].

Preliminary histopathological results performed on initial biopsies showed no correlation between endothelial cells marked via the ERGmarker, the number of vascular sections and C0 and C2 RGD uptake. Endothelial cell marking via the ERGmarker is an indirect expression of neovascularisation. The number of vascular sections per 3.15 cm² frame was counted. Further histopathological biopsies are necessary to confirm these results.

Patients with cHL had higher RGD uptake. cHL is mainly characterised by the presence of Reed Sternberg cells (RSCs), which are multinucleated neoplastic cells. RSCs are derived from B germinal centre and produce their own growth factors, Th2 cytokines and chemokines, creating a highly inflammatory infiltrate [13]. RSCs constitute only a minor component of the tumour (usually 1–3%); the majority of the malignancy is made up of a mixed inflammatory infiltrate variably composed of lymphocytes, eosinophils, fibroblasts, macrophages and plasma cells [14]. RSCs also express high levels of vascular endothelial growth factor (VEGF), which facilitates tumour progression [15]. The high RGD uptake in cHL compared to NHL could be due to αvβ3 overexpression in neovessels. An alternate explanation could be the presence of αvβ3 receptors on tumoural inflammatory cells. Complementary histopathological studies are being considered to verify these hypotheses.

SUVmax FDG was non-significant between responders and non-responders, because two patients presenting with DLBCL had an interim Deauville score of 4 and a final score of 2.

One of the major issues raised about RGD uptake is its ability to predict antiangiogenic drug response. Patients with DLBCL in our study had low RGD uptake. Seymour et al. [16] showed a lack of bevacizumab (anti-VEGF) efficacy in DLBCL when added to standard chemotherapy compared to chemotherapy alone. The low RGD uptake in DLBCL could therefore be linked to the lack of anti-angiogenic treatment efficiency. In contrast, patients with HL in our study had high RGD uptake. A pre-clinical study of xenografted Hodgkin lymphoma showed the efficiency of anti-angiogenic treatments [17]. No clinical data of antiangiogenic drugs in HL in humans are available. However, with the high RGD uptake in HL found in our study, the testing of anti-angiogenic treatments in refractory patients presenting with HL could be considered for further studies.

The increased C2 RGD uptake in NHL (Table 1) was not statically significant. Non-responders tended to have higher RGD uptake than responders on C2 PET (Fig. 5). Further studies are necessary to understand if the uptake is due to neoangiogenesis or an inflammatory process. Integrins are a class of heterodimeric cell surface adhesion receptors that are overexpressed on tumour endothelial cells during tumour angiogenesis, resulting in cancers that are more invasive, more migratory and better able to survive in different microenvironments. Among integrins, αvβ3 is highly expressed in tumours such as osteosarcomas, neuroblastomas, glioblastomas, malignant melanomas, and breast, lung and prostate carcinomas, but its expression is weak in most healthy organ systems (34).

The limit of our study is the small number of patients, but the patients did present with heterogeneous histological lymphoma subtypes.

Our study showed a trend of higher initial RGD uptake in patients presenting with cHL compared to NHL, and non-responders also had higher post-chemotherapy RGD uptake compared to responders. Issues raised about RGD uptake, particularly in cHL, are yet to be explored and need to be confirmed in a larger population.