To our knowledge, this work represents the first large study with a long follow-up dedicated to HCT-ASCT in patients with isolated VRL. Autologous stem cell transplantation with a thiotepa-based conditioning regimen has been the standard first-line consolidation treatment for PCNSL, according to two major randomized phase II studies (22–24). However, this treatment is still underutilized in PVRL, with limited data available in the literature. However, several studies from our team since the 2000s have shown the feasibility and effectiveness of HCT-ASCT for the treatment of VRL. Soussain et al. conducted a pilot study in 1996 involving five patients with refractory PVRL who underwent HCT, demonstrating its feasibility and showing encouraging results, with no cerebral relapse observed during the follow-up period (between 16 and 26 months) (25). Soussain et al. conducted another study in 2001, which further demonstrated the feasibility of this treatment and revealed promising outcomes (3-year overall survival of 63.7% in 22 patients, including 3 with relapsed VRL and 6 with isolated vitreoretinal relapse of PCNSL) (16). Subsequently, two additional studies were conducted: there was a 2-year OS rate of 69% after HCT-ASCT in a study of 43 relapsed patients including 5 with VRL (17), and a median OS of 86 months after HCT-ASCT in 19 patients with relapsed VRL (26). More recently, Lee et al. described two patients with VRL treated with first-line HCT-ASCT who achieved long-term survival with follow-up durations of 7 and 8 years, respectively(27).
The results observed in our study were generally favourable (median OS, BFS, and PFS from HCT-ASCT of 92, 113 and 96 months, respectively). The PFS from HCT-ASCT was much longer than the PFS from initial diagnosis. These results represent a better prognosis than that usually described in the literature for PVRL (median overall survival from initial diagnosis ranging from 37 to 75 months in the main studies, with heterogeneous treatments ranging from local treatments to systemic treatment mainly based on high-dose methotrexate or to combined systemic and local treatments) (2, 10, 12) as well as for isolated ocular relapses of PCNSL (median overall survival from the date of isolated ocular relapse: 57.1 months) (19). We can reasonably imagine that most of the patients who were still disease-free 5 years after HCT-ASCT were definitely cured with this strategy. However, these results must be interpreted cautiously, especially given the retrospective nature of this study, with a high proportion of young patients having a good KPS, which is a factor well known to be associated with a better prognosis in PCNSL (5, 28, 29).
Most of the patients in our study received a thiotepa-based HCT. Due to the limited number of patients, it was difficult to compare the effectiveness of the various regimens or of thiotepa-based HCT versus BEAM. However, with 4 of 6 patients who relapsed following BEAM and given the poor results of this regimen in PCNSL cases (18, 30), our recommendation would be to prefer thiotepa-based regimens.
Despite the favourable survival outcomes, the post-HCT-ASCT relapse rate, particularly the cerebral relapse rate, remains high, both in patients with PVRL and in patients with PCNSL at initial diagnosis (45% at 5 years, 45% in patients with initial PVRL, 44% in patients with initial PCNSL). An important issue that could partly explain this high rate of relapse is the difficulty in adequately evaluating the quality of response in VRL patients. Currently, according to the IPCG criteria (20), the response to treatment is assessed by a rather subjective clinical ophthalmological evaluation. There is no measurable mass in VRL, and many patients maintain residual abnormalities that may be either only a scar or the persistence of an active disease (31). We could therefore hypothesize that some patients from our cohort did not exhibit a good response at the time of HCT-ASCT, which is a well-known adverse prognostic factor regarding the efficacy of HCT-ASCT in PCNSL (32). The systematic assessment of IL-10 in the aqueous humour might be an interesting tool (1). This marker may have a greater sensitivity than clinical examination in detecting relapses early, although further research is needed to confirm this issue. In our cohort, we had limited data, with only 13 patients whose IL-10 levels were measured prior to HCT-ACST. However, three of five patients with elevated IL-10 levels at the time of HCT-ASCT subsequently relapsed.
The majority of the patients in our series received HCT-ASCT in second line or subsequent lines of treatment. In PCNSL patients, the relapse rate is low following HCT-ASCT performed as a 1st-line treatment (5-year PFS of approximately 75%) (18) but is significantly higher in the setting of relapse, with rates close to those observed in our series (2nd -line treatment: 5-year PFS of 48% in PCNSL patients (18) vs. 44% in VRL patients in our series). In our cohort, only three patients received first-line HCT-ASCT, two of whom never relapsed after 5 and 9 years of follow-up. However, this represents a cohort that was too small to draw any adequate conclusions. Considering the similarity of the results observed in PCNSL and in PVRL, we could hypothesize that the results of HCT-ASCT in the 1st-line treatment of PVRL would be better than those in the setting of relapse. In light of the poor long-term prognosis of this disease, the use of HCT-ASCT as a 1st-line treatment for VRL should be discussed, especially considering the fact that most PVRL patients experience relapse after "conventional" first-line treatments (with a 5-year PFS of approximately 25% (2, 12)). Given the good KPS of most PVRL patients, this strategy could be used in many of these individuals up to a fairly advanced age, as is done in select elderly patients with PCNSL (22, 33, 34).
The toxicity profile observed was close to what is usually observed in other studies involving HCT-ASCT (18). There was a 3% rate of toxicity-related mortality, which is slightly lower than the 5 to 10% range reported for PCNSL (17, 23, 26, 35). Even if this toxic mortality rate appears to be acceptable, HCT-ASCT remains a toxic treatment that is difficult for many patients to tolerate, and patients should be aware of the limited but real risk of toxic death. Optimizing induction treatments for PVRL, potentially through the incorporation of novel drugs (imids, iBTK, etc.) (36–38), could limit the need for HCT-ASCT by increasing the rate of complete response. We also currently lack molecular prognostic factors that could indicate whether patients belong to a group with a higher or a lower risk of relapse. Several recent studies have revealed promising prognostic molecular factors in PCNSL cases (39, 40), but reliable tools for identifying these factors in patients with VRLs are lacking. There might also be a role for the use of CAR T-cells, which has yet to be determined. Indeed, in systemic lymphomas, therapeutic strategies using CAR T-cells compare favourably to standard strategies, including HCT-ASCT (41, 42). Promising preliminary data on CNS lymphoma treated with CAR T-cells have been published in recent years, but there are limited data regarding ophthalmological involvement (43, 44).