The incidence of grade 3 to 4 OM differed depending on conditioning regimen. A lower incidence was observed with FluTBI mini, FluCyTBI (PTCy), FluCy and MelFluTBI (PTCy). This indicates that 2Gy TBI, fludarabine and high dose cyclophosphamide are unlikely to cause severe OM. Therefore, the higher incidence of grade 3 to 4 OM in CyTBI and FluTBI is mostly due to myeloablative high dose TBI and this is considered as radiation-induced OM. When comparing similar regimens, CyTBI vs FluTBI (PTCy), FluMel vs MelFluTBI (PTCy) and FluCy vs FluCyTBI (PTCy), the use of MTX as an immunosuppressant appears to be a risk factor compared to PTCy. Melphalan is known to cause OM. However, given the relatively low incidence of grade 3 to 4 OM in patients who received HDM autologous and Mel FluTBI (PTCy), it appears that melphalan is a risk factor when combined with MTX or in BEAM rather than the dose of melphalan itself. Among the commonly used regimens, the incidence of grade 3 to 4 OM is highest in CyTBI, then FluMel and BEAM followed by HDM.
A higher incidence of grade 3 to 4 OM has been previously observed with myeloablative conditioning compared to reduced intensity conditioning utilizing busulfan- based regimens [13]. In this study, grade 3 to 4 OM was observed in 45% of patients receiving a myeloablative regimen, which is lower than grade 3 to 4 OM experienced in our CyTBI cohort (71%) and comparable to our FluTBI and FluMel cohorts (46% and 43%). A large systematic review that included various regimens did not identify the different incidence or severity of OM between myeloablative and reduced-intensity conditionings [14]. The risk of OM appears to be determined by specific drug combinations and use of MTX-based immunosuppression (in allogeneic HPCT) rather than the conditioning intensity alone. Methotrexate, as an immunosuppressant, has previously been reported as a risk factor for OM [14, 16, 9]. Our study collaborates this, and suggested superiority of PTCy in terms of OM prevention. Although melphalan dose has been shown to predict OM in one study [17], in our experience, melphalan doses were not relevant to the incidence of grade 3 to 4 or grade 2 to 4 OM. Use of melphalan in combination with other agents in conditioning (carmustine and etoposide) or GVHD prevention (MTX) appeared to be needed for development of OM.
The secondary outcomes showed similar trends with the incidence on grade 3 to 4 OM. The incidence and duration of grade 2 to 4 OM were relatively lower in FluTBI (PTCy) and higher in FluCyTBI (PTCy) compared to the primary outcome. Small sample size in these regimens may have affected the results. PCA use was corelated to the incidence of grade 3 to 4 OM. The use of TPN was lower in BEAM and HDM autologous patients, which reflected the unit’s practice of limiting TPN in autologous patients. There are conflicting results in regard to the association between OM and taste disturbance [21, 22]. In our study, the incidence of taste disturbance was high regardless of the regimen and it was not associated with severe OM. Taste disturbance in the early post HSCT period is most likely due to direct conditioning-related oral mucosal and salivary gland injury [4]. Taste disturbance may present with lower level of toxicities compared to OM. Taste dysfunction affects patients’ eating and wellbeing after the treatment, and potentially TPN use after HSCT.
There are other reported risk factors than HSCT regimens, including younger age, female gender and lower BMI. However, it is difficult to assess these factors independently. For example, younger patients may have higher risks because they are more likely to have myeloablative regimens. In our study, 197 (42%) patients underwent FluMel allogeneic HSCT. In this population, the incidence of grade 3 to 4 OM, use of TPN and PCA were higher in female patients. As all patients underwent the same treatment, the effect of treatment-related risk factors was minimised. Female gender has been consistently reported as a patient-related risk factor [5–7]. This could be because females tend to be smaller and if chemotherapy is calculated with body surface area they generally receive a higher dose per kg. Another explanation maybe the negative effects of female sex hormones as pointed out in a recent study [5]. The gender difference in the incidence of grade 2 to 4 OM was not statistically significant. Grade 2 to 4 OM is determined by the presence of ulcers and more objective than other measures. Therefore, the gender differences in G3-4 OM, TPN and PCA uses may be interpreted partly as gender difference of pain perception and oral intake affected by pain. A large study demonstrated that women report increased clinical pain compared to men [23].
In our study, median BMI and age did not vary across the patients with different OM grades after FluMel conditioning. This finding is at variance with the results of a study a published by Shouval et al in 2019 [9]. This study involved a heterogenic patient cohort receiving a wide range of conditioning regimens, and their multivariate analysis did not show age and BMI as significant risk factors.
In our study, it was confirmed that severe OM is associated with HSCT conditioning and immunosuppressant regimens containing myeloablative TBI, methotrexate and melphalan in combination with methotrexate or in BEAM. Currently the only FDA approved prophylaxis of OM is palifermin. Palifermin appears to be effective only for TBI-induced OM and there are associated costs and accessibility issues with this medication. Clearly, studies to identify more accessible and inexpensive interventions are required. Challenges in OM studies include heterogenicity of patients, disease and treatment. Based on our study and other published studies, when conducting a clinical trial, it is appropriate to stratify randomisation according to conditioning and immunosuppressant regimens if the study includes patients undergoing different regimens.
Our study had some limitations. First of all, this was a retrospective study using daily OM assessment by nurses. In a busy HSCT unit, we observed their OM gradings were not always accurate [24]. Due to the retrospective nature of the study, we had to rely on their observations. Secondly, HSCT regimens were dramatically changed towards the end of study period from matched unrelated donor allograft regimen to haploidentical regimens due to donor unavailability associated with COVID-19 pandemic. As a result, there were some haploidentical regimens with small sample size. On the other hand, these regimens provided possible superiority of PTCy to methotrexate to prevent OM.
In summary, the incidence of severe OM was largely different depending on conditioning and immunosuppression regimens. Female gender was found to be a risk factor. Taste disturbance was common regardless of the regimens. In clinical trials including patients undergoing different HSCT regimens, it is best to stratify randomisation according to the regimens.