Four studies were included in the treatment comparison between Lonca and Pola + BR. LOTIS-2 for Lonca and the GO29365 extension study, the COTA database, and Dal et al. 2023 for Pola + BR.

LOTIS-2 is a Phase 2, multicenter, open-label, single-arm study of the efficacy and safety of Lonca used as monotherapy in 145 patients with R/R DLBCL after 2 or more systemic therapies.

The GO29365 extension study [28] comprises a multicenter, open-label, randomized, Phase 1b/2 study of Pola + BR for patients with R/R DLBCL and at least one prior line of therapy. In the Phase 2 randomized phase, 80 patients were randomly assigned in a 1:1 ratio to the Pola + BR and BR treatment arms. A further 106 patients were enrolled into a single-arm extension study and received Pola + BR.

The COTA database study is a real-world study of adults diagnosed with DLBCL, treated with Pola + BR or Tafa + Len and identified from COTA electronic medical record (EMR) data [40]. Outcomes were analyzed for 79 patients treated with Pola + BR and aggregate data were extracted from the conference poster for the study.

Dal et al. 2023 consists of observational, retrospective, RWE from 32 hematology centers across Türkiye in patients with R/R DLBCL who had received at least two prior treatments, with progression or intolerance with their last treatment [30].

Relative Efficacy: LOTIS-2 versus Individual Comparator StudyGO29365 Extension Study

Patients with transformed lymphoma were excluded from the GO29365 RCT and extension study [28]. Therefore, to align study inclusion criteria as closely as possible, 29 patients with transformed lymphoma were excluded from the LOTIS-2 dataset. Although the majority of patients were third- or later-line (102/152), the GO29365 extension study enrolled patients in second- or later-line treatment, and baseline characteristics were only available for the overall population, with subgroup data available for all efficacy outcomes of interest for patients at third- or later-lines.

For both pre- and post-matching comparisons, ORR was very similar for Lonca compared with Pola + BR in the relevant population, with an OR of 0.92 (95% CI 0.52, 1.62) after characteristic matching (Table 3; see also Table S7). However, the results for CR significantly favored Pola + BR, with an OR of 0.46 (95% CI 0.26, 0.81) pre-matching and an OR of 0.47 (95% CI 0.25, 0.89) post-matching.

Table 3 Efficacy results for the MAIC comparisons between Lonca and Pola + BR (OR > 1.0 and HR < 1.0 favor Lonca)

Despite the difference in CR, no significant differences were observed in either PFS or OS between treatments (Table 3). However, only median PFS and median OS with 95% CI and number of events were reported for the third- or later-line subgroup in the GO29365 extension study. This led to a HR (95% CI) estimation for Lonca versus Pola + BR using point estimates for median survival and the number of events. The Kaplan–Meier (KM) plots for PFS for patients receiving Lonca for the unadjusted and MAIC weighted patient data are shown in Figure S5. Additional PFS results are provided in Table S8.

OS was similar between the two treatments (HR close to 1.0; Table 3), with no treatment significantly favored over the other. The KM plots for OS for patients receiving Lonca for the unadjusted and weighted patient data are shown in Figure S6. Additional OS results are provided in Table S9.

COTA Database

All patients from LOTIS-2 were included in the IPD dataset for the comparison with third- or later-line subgroup of Pola + BR patients from the COTA database [40]. There was no evidence of a significant difference between Lonca and Pola + BR in the odds of overall response, with the 95% CI for the OR crossing 1.0 in all estimations based on both unweighted and weighted data (Table 3; Table S11). For CR, the point estimate of the OR numerically favored Lonca, although the result was not statistically significant.

PFS with Lonca was significantly better than Pola + BR, both pre- and post-matching, with an adjusted HR of 0.66 (95% CI 0.44, 0.97) (Table 3). However, this estimate should be treated with caution as a visual inspection of the KM curves suggests that the proportional hazards assumption (PHA) may be violated. KM plots for PFS for patients receiving Lonca for the unadjusted and weighted patient data compared with those receiving Pola + BR are shown in Figure S7.

OS was numerically improved with Lonca compared with Pola + BR, but the difference was not statistically significant: adjusted HR 0.69 (95% CI 0.46, 1.04) (Table 3). As with PFS, it was noted that the PHA is likely to be violated, due to the crossover of the curves early on during follow-up. KM plots for OS for patients receiving Lonca for the unadjusted and weighted patient data compared with those receiving Pola + BR are shown in Figure S8.

Dal et al. 2023

All patients from LOTIS-2, except two, were included in the IPD dataset for the comparison with Pola + BR from the Dal et al. 2023 study [30]. The two excluded patients were older than the eldest patients reported in the range of ages for Dal et al. 2023 (> 84 years), and were excluded in order to better align the overall characteristics of the datasets before the matching process was conducted. Response was measured in both studies using the 2014 Lugano classification [34], and the OR estimates for ORR and CR, pre- and post-weighting, showed no significant differences when comparing Lonca with Pola + BR (Table S15).

No data were reported for PFS. OS was improved for patients receiving Lonca compared with Pola + BR (HR < 1.0; Table S16). In the unweighted estimate and weighted bootstrap estimate, Lonca offered significantly longer survival than Pola + BR [unweighted HR for Lonca vs. Pola + BR 0.58 (95% CI 0.42, 0.81) and weighted HR 0.65 (95% CI 0.42, 0.98)]. The KM plots for OS for patients receiving Lonca for the unadjusted and weighted patient data compared with those receiving Pola + BR are shown in Figure S9.

Extranodal disease at baseline was also reported for this study, although, as this characteristic was not identified as a key prognostic factor, it was not included in the base case analyses. In a sensitivity analysis, the percentage of patients with extranodal disease was included as a matching variable for the key outcome of interest, OS. Including this additional variable for matching reduced the effective sample size further (Table S17). The results for OS were consistent with the base case and OS was significantly improved with Lonca compared with Pola + BR across all unadjusted and weighted estimates (HR < 1.0) (Table S18).

Relative Efficacy: Meta-analyses

In addition to the calculations made for each individual comparator study, overall comparisons between Lonca and Pola + BR were of interest for generalizable and meaningful overall conclusions.

For the meta-analysis of response outcomes, there was no significant difference between the two treatments for ORR or CR (Fig. 2). Although there were differences in study design, the estimates of ORR were very consistent with a lot of overlap between the 95% CIs from the individual MAIC outputs. The ORs were identical between fixed and random effects models for ORR as there was no heterogeneity identified between the MAIC outputs (Fig. 2a). For CR, the random effects model was also considered the most appropriate, particularly as substantial heterogeneity was noted between the individual study MAICs (for example, I2 = 59%; Fig. 2b).

Fig. 2

Meta-analysis outputs for response and survival outcomes—Lonca matched to Pola + BR: a ORR; b CR; c OS; d PFS. For response outcomes, OR > 1.0 favors Lonca (greater number of response events is preferred), and for survival outcomes, HR < 1.0 favors Lonca (smaller hazard of experiencing an event is preferred). CI confidence interval, CR complete response, ext extension, HR hazard ratio, Lonca loncastuximab tesirine, OR odds ratio, ORR overall response rate, OS overall survival, PFS progression-free survival, Pola + BR polatuzumab plus bendamustine plus rituximab

Although there was a numerical advantage for Pola + BR in the odds of achieving response outcomes (OR < 1.0 for Lonca vs. Pola + BR), this did not translate into an OS benefit (Fig. 2). In fact, although not statistically significant, Lonca was numerically favored for OS (HR < 1.0), with an estimated decrease in the hazard of death for Lonca of 21% compared with the hazard of death in the Pola + BR-treated group [HR 0.79 (95% CI 0.60, 1.05)] (Fig. 2c). The hazard rates for PFS were more similar with no evidence of a difference between Lonca compared with Pola + BR [HR 0.91 (95% CI 0.49, 1.69)] (Fig. 2d).

For the OS meta-analysis, moderate heterogeneity was identified (I2 = 32%); while for PFS, considerable heterogeneity was identified (I2 = 82%).

Relative Safety and Tolerability

Safety and tolerability outcomes were not reported in the COTA database study and there was limited reporting in the Dal et al. 2023 [30] study. The GO29365 extension study reported relevant safety data to inform a comparison between Lonca and Pola + BR.

GO29365 Extension Study

As with the efficacy outcome comparisons, 116 patients from LOTIS-2 were included in the comparison of safety for the GO29365 extension study. Analyses were adjusted for age and ECOG PS and were conducted for outcomes available from the GO29365 extension study, including treatment discontinuation due to AEs as primary reason for discontinuation (related or emergent, not specified); fatal AEs; Grade 3–4: neutropenia, thrombocytopenia, anemia, infections and infestations, and any Grade 3–4 AEs; SAEs: febrile neutropenia, sepsis, pneumonia, pyrexia, and any SAE.

Follow-up time and treatment duration differed between the trials. For LOTIS-2, follow-up was 3 years with a median treatment duration of 45 days (range: 1–569 days). In the GO29365 extension study, follow-up was 48.9 months in the randomized arm and 15.2 months for the extension cohort with a median treatment exposure duration of 97 days (range: 0–212 days) across the entire cohort. However, an exposure-adjusted analysis was not deemed appropriate (see Supplementary Information) and using standard binary outcome comparisons was considered to be a conservative approach for Lonca.

In the population-adjusted comparison, compared with Pola + BR, Lonca was associated with significantly lower odds of Grade 3–4 infections and infestations [OR 0.34 (95% CI 0.15, 0.78); Table 4], and significantly lower odds for any SAE [OR 0.51 (95% CI 0.31, 0.86)] and SAE of febrile neutropenia [OR 0.24 (95% CI 0.07, 0.89)], pneumonia [OR 0.10 (95% CI 0.01, 0.78)] and pyrexia [OR 0.17 (95% CI 0.04, 0.82)].

Table 4 Comparison of safety outcomes in the weighted population: Lonca (LOTIS-2) versus Pola + BR (GO29365 extension study)

For all other AE outcomes analyzed, including any AE of Grade 3–4, Grade 3–4 anemia, neutropenia, and thrombocytopenia, and fatal AEs and sepsis, the point estimates for the ORs numerically favored Lonca (OR < 1.0); however, these were not statistically significant differences. Discontinuations due to AEs were similar between treatments. There was little difference between the numerical values of the OR in the unadjusted and adjusted comparisons (see Table S19 for unadjusted estimates).