In our network meta-analysis evaluating first-line treatments for diverse solid tumors, we found that patients treated with Chemo-IO experienced a higher rate of treatment discontinuation due to TRAEs than those receiving mono-IO.
The advent of immunotherapy has ushered in a new era in cancer treatment. Starting with mono-IO, the field has since evolved to incorporate more complex regimens, such as IO-IO combinations and Chemo-IO, which have shown improved survival outcomes over Chemo-only or mono-IO [2, 6, 7, 12]. These advancements have led to approvals from the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA) for various cancer treatment settings [7, 8, 13, 14].
However, due to the additive nature of adverse effects, Chemo-IO is generally associated with a higher incidence of TRAEs compared to mono-IO [8, 9]. TRAEs not only cause direct harm but can also potentially lead to negative impacts on prognosis by prompting treatment discontinuation. Consistent with these findings, our study observed a significantly increased rate of TRAE-induced discontinuation in the Chemo-IO group compared to the mono-IO group (RR 2.68, 95% CI 1.98–3.63) across various cancer types. Similarly, among NSCLC patients, Chemo-IO also demonstrated an elevated rate of discontinuation due to TRAEs compared with mono-IO (RR 2.93, 95% CI 1.67–5.14).
While the optimal duration for IO treatment continues to be a topic of ongoing debate, several trials, including KEYNOTE-189 [15], KEYNOTE-010 [16], and CheckMate-153 [17], suggest the possibility of poor survival outcomes or disease progression after discontinuing IO treatment following one or two years of administration [18]. Limited research has managed to follow patients who discontinue treatment due to AEs. However, one retrospective study that examined the clinical outcomes of patients who discontinued IO treatment due to immune-related AEs found that 20% of patients experienced disease progression within six months of discontinuation and 10% of patients died [19]. This evidence further emphasizes the potential impact of treatment discontinuation on patient outcomes.
In susceptible populations, such as elderly patients or those with a poor Eastern Cooperative Oncology Group performance status (ECOG PS), the administration of Chemo-IO could potentially be associated with a poorer prognosis due to an increased incidence of severe TRAEs and subsequent treatment discontinuation owing to these AEs [20]. Moreover, even with Chemo-only regimens, certain patient subgroups such as those of advanced age, poor performance status, individuals with anemia, impaired renal function, hearing impairment, or history of falls, are known to have an increased risk of toxicity [21]. This heightened toxicity could potentially lead to treatment discontinuation in these subgroups.
For high-risk patients predicted to cease treatment due to TRAEs, a treatment strategy focusing on mono-IO might offer a lower discontinuation rate, thus improving treatment continuity. Furthermore, considering that several studies have reported comparable effects of IO to Chemo-IO in certain subgroups [14, 22], there is a pressing need for further research into potential biomarkers or clinical factors that can aid in identifying patients who would benefit sufficiently from mono-IO instead of Chemo-IO.
Several predictive biomarkers, including PD-L1, tumor mutation burden (TMB), and tumor-infiltrating lymphocytes (TILs), have been proposed to predict the efficacy of IO, but their roles remain inconclusive [3, 4, 8, 23]. Consequently, novel approaches are being explored to improve prediction accuracy, including the application of artificial intelligence (AI). Some studies have even reported that AI-assisted methods, such as the evaluation of pretreatment contrast-enhanced CT images, PD-L1 expression, or the spatial analysis of TIL, can yield better predictions of survival outcomes in NSCLC patients undergoing IO treatment [24–26]. Pursuing such research avenues may potentially enhance our capacity to provide personalized treatment strategies to individual patients, thereby potentially improving patient outcomes.
Limitations
While this meta-analysis provides valuable insights, it does have certain limitations. First, the potential for bias due to confounding effects could vary based on the types of immunotherapies used across the different regimens and the diversity in types of solid tumors. However, due to the lack of sufficient data from the RCTs, we did not differentiate among these agents in our analysis. Second, our study is primarily based on published literature and clinical trial results, which could potentially lead to publication bias, as studies with negative results are less likely to be published. Third, the duration of follow-up in each RCT differed, potentially contributing to increased rates of discontinuation due to AEs. Last, we did not have access to individual patient data, which restricted us from performing subgroup analysis based on various clinical factors.