Targeted therapies in retinoblastoma: GD2-directed immunotherapy following autologous stem cell transplantation and evaluation of alternative target B7-H3

Retinoblastoma (RB) is the most common malignant eye-tumor in children [1]. Stage-adjusted therapy strategies ranging from local therapies, systemic chemotherapy, radiation to high-dose chemotherapy with autologous stem cell transplantation (ASCT) result in cure rates as high as 95% [2, 3]. In localized stages, the eye or even vision can be preserved, whereas management of advanced local or metastasized disease remains challenging [4, 5]. Interestingly the disialoganglioside GD2 and B7-H3 may represent potential therapeutic targets in these tumors:

Individual treatment attempts using Anti-GD2 mAb in patients with advanced retinoblastoma were undertaken by different treatment centers [18, personal communication]. We here describe a cohort of four patients with metastatic relapse of retinoblastoma, who were treated with the Anti-GD2 monoclonal antibody Dinutuximab beta (ch14.18/CHO) following autologous stem cell transplantation in an individual compassionate use attempt in three centers in Germany.

Altogether 4 patients with metastatic relapse of retinoblastoma, who were treated with Dinutuximab beta between 2018 and 2021, were identified in the RB-registry. Treating centers were Essen, Tuebingen and Freiburg. Patient details are summarized in Table 1. While patient 4 was affected by sporadic retinoblastoma without germ line alteration of RB1, the others harbored a heterozygous pathogenic variant in the RB1 gene in constitutional DNA. These alterations in constitutional DNA, which represent a predisposition for the development of Retinoblastoma, did not manifest in the families before or represent a de novo mutation, respectively, and diagnosis was made after clinical manifestation of bilateral Retinoblastoma at a young age between 4 and 15 months. No extraocular, intracranial and metastatic disease was found at initial presentation.

Despite stage-adjusted therapy, all patients suffered from one or more relapse or progression, respectively, resulting eventually in metastasized Stage IV disease according IRSS. Multimodal therapy in all patients consisted of two to six treatment lines, including at least two different schemes of systemic chemotherapy, enucleation and radiotherapy. In patients with bilateral disease eye-preserving treatment of at least one eye was attempted with local therapies like cryocoagulation, brachytherapy, selective intraarterial chemotherapy and intravitreal chemotherapy. Nevertheless, sequential enucleation of the second eye was inevitable in two patients. Finally, following rescue therapy all patients received consolidating high-dose chemotherapy with subsequent ASCT.

The risk of renewed recurrence was considered to be high in all of the reported patients by the respective treating center. Evaluation of tumor tissue from the primary tumor or metastasis for therapeutic targets revealed an expression of GD2 on tumor cells (Fig. 1B). Therefore, analogous to high-risk NB protocols, 5 cycles of post-consolidation immunotherapy with Dinutuximab beta were applied on a compassionate use basis after informed consent for off-label therapy. Dose per cycle was 100mg/m² over 5 days (patient 3) or 10 days (patients 1, 2, 4), respectively, according to treatment centers standard every 5 weeks. Supportive medication was given as described [23] including Morphine, Pregabalin or Gabapentin following local standards.

While treatment with mAb Dinutuximab beta was tolerated without serious side effects in patient 1 and 4, the first antibody cycle had to be interrupted in patient 3 due to capillary leakage with respiratory insufficiency on the second day. Symptoms resolved fast and the course was recommenced with a lower infusion rate. Subsequent courses were tolerated satisfiable.

Side effects in patient 2 were more profound with respiratory insufficiency, capillary leakage, arterial hypotony, urticaria and bronchial obstruction. The first cycle was reduced by 2 of 10 days, the second had to be interrupted. Eventually therapy had to be permanently stopped after anaphylactic reaction at the third cycle.

In vitro investigations showed high expression of GD2 on all examined RB cell lines (Fig. 1A). Cytotoxicity assays using PBMCs of patient 3, revealed their capability to specifically lyse the retinoblastoma cell line RBL-30 by ADCC when combined with ch14.18/CHO. Compared to patient’s PBMCs alone, target cell lysis was significantly increased when ch14.18/CHO was present (Fig. 2A). ch14.18/CHO main effector mechanism also relies on serum complement. Thus, we additionally tested patient’s serum for its capability to induce target cell lysis by CDC. Serum was taken 12 h after anti-GD2 mAb application on day one of the 5th cycle. Serum in combination with patient’s PBMCs but also serum alone significantly increased RB target cell lysis compared to control groups containing heat-inactivated or no serum at all serum. This clearly demonstrates the presence and the cytolytic activity of ch14.18/CHO in patient’s serum during the course of mAb therapy (Fig. 2B). Taken together, these data show the functionality of patient’s PBMCs to induce ADCC in combination with ch14.18/CHO and that mAb is contained in sufficient amounts within patient’s serum to mediate CDC in vitro.

Follow-up ranged from 11 to 45 months. Since patient 2 and 4 received Dinutuximab beta in complete remission (CR), response evaluation was restricted to the remaining two patients. In patient 1 paravertebral metastasis was no longer detectable after resection and systemic chemotherapy, whereas metastasis at the left mandibula persisted. In patient 3 the residual tumor mass in the left orbita remained after ASCT and was evaluable for further treatment response. Regression of target lesions could be observed in both patients, leading to CR in patient 1 (Fig. 3A–C) and tumor regression in patient 3 (Fig. 3D–E).

Despite initial response in patient 3 to antibody treatment, patient 3 and 4 suffered from a subsequent intracranial relapse 12 and 10 months after the start of GD-2 therapy, respectively, resulting in an Event Free Survival of 50% after 2 years. Renewed biopsy of metastasis in patient 3 showed a loss of GD2 expression (Fig. 1B–C). Instead, a strong expression of B7-H3 was found (Fig. 1D). B7-H3 is known to be expressed on a platitude of solid tumor and a clear expression could also be confirmed on all of our evaluated RB cell lines (Fig. 1A). Compared to GD-2, B7-H3 is likewise expressed on cell lines RBL-30 and WERI-RB-1. In case of Y79 though, B7-H3 is expressed on a rather low level when compared to GD-2. Using a fully humanized anti-B7-H3 mAb (huA5-B7-H3), patient’s PBMCs and PBMCs from a healthy donor mediated effective cytolysis against RBL-30 which, however, was less pronounced when compared to ch14.18 (Fig. 2A). As huA5-B7-H3 is not yet approved at the stage of a new relapse, palliative radiotherapy was applied and the patient deceased six months later.

In patient 2, an alveolar Rhabdomyosarcoma (aRMS) developed 5 years after diagnosis of bilateral retinoblastoma, but before its metastatic relapse occurred. Initial treatment of retinoblastoma in this patient consisted of systemic chemotherapy, photon irradiation of the left eye in Schipper technique and sequential enucleation of the right and later the left eye. The aRMS was localized at right Os zygomaticum/Maxilla [N+, M0], assumably in close proximity of the previous irradiation field. The aRMS was successful treated according corresponding protocols with chemotherapy, irradiation and resection.

Experience in treatment of neuroblastoma with Dinutuximab beta draw interest in this therapy against other tumors expressing GD2, among others retinoblastoma [7]. By the time of our studies, immunotherapy using anti-GD2 antibodies in retinoblastoma was reported simultaneously by other treatment centers [18]. Two studies using I131-labeled Anti-GD2 antibody 3F8 in GD2 positive tumors permitted, among other entities, the inclusion of patients with retinoblastoma [24, 25]. To our knowledge, there are currently no prospective studies above this, evaluating anti-GD2 mAb in retinoblastoma specifically. The rarity of high-risk stage IV retinoblastoma impedes the realization of such studies.

Monoclonal antibodies like Dintuximab beta are supposed not to penetrate the healthy blood–brain barrier (BBB). When drug levels of Db in cerebrospinal fluid were determined in three patients with acute neurotoxic side effects of the immunotherapy, no relevant levels were measured [26]. Actually, relapse after Db immunotherapy occurred in both of our patients intracranial. If this can be related to insufficient penetration of Db into the central nervous system or reflects common route of RB metastasis can´t be answered, especially with respect to GD2 negative relapse in patient 3. On the other hand, some findings indicate, that immunotherapeutic approaches with anti-GD2 antibodies can be a promising therapeutic option for GD2-expressing intracerebral or intraretinal tumors and metastases. Disrupted blood-retinal barrier in retinoblastomas [13] and the increased BBB-permeability caused by NB/RB metastasis and by irradiation, can allow effector cells and mAbs to exert effects in CNS tumors. Furthermore, the murine antibody 14.G2a was found to bind to the cerebellum [27], and the chimeric antibody m3F8 has been reported to cause disruption to the BBB itself [28].

Our retrospective analysis has several limitations. The cohort is very small without any control group. Patient 2 did not receive the full treatment of Db due to side effects. Patients received the immunotherapy following ASCT ± radiation and the effects of the different treatment modalities can´t be completely distinguished. Thus, the analysis does not allow to draw firm conclusions on the efficacy of the approach. However, anti-GD2-directed immunotherapy resulted in a response in both patients with evaluable residual tumor seen in MRI. These results, in addition to our in vitro assays, encourage further investigation in high-risk retinoblastoma, although prospective randomized studies are unlikely to occur due to the small number of expected eligible patients.

Four years after diagnosis of retinoblastoma and its treatment, including radiation, secondary alveolar Rhabdomyosarcoma (aRMS) developed in patient 2. Heterozygous pathogenic variants in RB1 in constitutional DNA represents not only a risk factor for the development of retinoblastoma but also other subsequent primary malignancies (SPM) particularly soft tissue sarcoma and osteosarcoma [29, 30]. Risk for secondary soft tissue sarcoma is even higher after external beam radiotherapy (EBRT) [29] and soft tissue sarcoma are more likely to develop on the head within the radiation field [31]. This effect was found to be dose dependent [30] and age related [32]. To account for these relations, the use of radiotherapy was confined in favor of chemotherapy [2]. Especially in those patients with hereditary retinoblastoma, GD2-directed immunotherapy might be of interest to substitute or delay radiotherapy.

Despite intense pretherapy in all patients, safety and tolerability of the immunotherapy was acceptable and comparable to the use of Dinutuximab beta in neuroblastoma patients [33]. Only in patient 2 therapy had to be terminated definitely due to anaphylaxis, which is a known common (rate from ≥ 1/100 to < 1/10) side effect [34].

Furthermore, no cumulative toxicity or carcinogenic potential of anti-GD2-directed immunotherapy has been reported and immunotherapy might be repeated in case GD2-positive metachronous tumors.

In patient 3 a GD2-negative relapse occurred. Loss of GD2 is a rare event in neuroblastoma following anti-GD2 immunotherapy [21, 35] and can be partly explained by reduced antigen production or by internalization of antigen/antibody complexes [36, 37]. In our patient, theoretically a separate GD2-negative retinoblastoma could have developed on the basis of the RB1 germline mutation, independent from the immunotherapy. Metachronous tumors and secondary malignancies are not uncommon in patients with hereditary retinoblastoma [38, 39]. Since both eyes were enucleated at this stage and no retinal tissue left, from which a new retinoblastoma could have developed, we regard this unlikely.

Thus, we suppose that in our patient a GD2 negative retinoblastoma occurred, which escaped the Dinutuximab therapy. Indeed, in vitro studies evaluating cytotoxicity of Anti-GD2 CAR T-cells in RB cell-lines showed a downregulation of GD2 [15]. However, if a loss of GD2 occurs more often in retinoblastoma than in neuroblastoma can´t be answered here. The escape of GD2 negative clones might be prevented by directing immunotherapy against different targets subsequently or in parallel. In case of GD2-negative tumors or GD2 loss, B7-H3 might serve as a potential alternative target. Several anti-B7-H3 therapies are currently investigated [40, 41].

Expression of B7-H3 both on RB cell lines and the tumor of patient 3 was shown. Using patient’s PBMCs and an anti-B7-H3 mAb, a clear ADCC effect against the RB cell line RBL-30 was demonstrated.

While survival for retinoblastoma in high-resource countries generally is high, there is a great global disparity [42]. Presentation in low-income countries (LIC) is usually at late stage [43, 44], the disease seems to be more aggressive and access to therapy is limited [45]. Estimated mortality of 40–80% [42, 45] highlights the need of alternative therapies in these countries. Despite anti-GD2 immunotherapy is currently high priced, the therapy might be feasible in low-income setting, particularly regarding less pronounced cytopenia.

In summary, we rate Anti-GD-2 immunotherapy a promising alternative in treatment of retinoblastoma patients. Especially in patient with disposition for retinoblastoma saving chemotherapy or radiation is most important due to their carcinogenic effects. Also, therapeutic options for advanced stages of retinoblastoma or relapsed disease are needed.