Trilaciclib prior to chemotherapy and atezolizumab in patients with newly diagnosed extensive‐stage small cell lung cancer: A multicentre, randomised, double‐blind, placebo‐controlled Phase II trial

Abstract Trilaciclib is an intravenous CDK4/6 inhibitor administered prior to chemotherapy to preserve haematopoietic stem and progenitor cells and immune system function from chemotherapy‐induced damage (myelopreservation). The effects of administering trilaciclib prior to carboplatin, etoposide and atezolizumab (E/P/A) were evaluated in a randomised, double‐blind, placebo‐controlled Phase II study in patients with newly diagnosed extensive‐stage small cell lung cancer (ES‐SCLC) (NCT03041311). The primary endpoints were duration of severe neutropenia (SN; defined as absolute neutrophil count <0.5 × 109 cells per L) in Cycle 1 and occurrence of SN during the treatment period. Other endpoints were prespecified to assess the effects of trilaciclib on additional measures of myelopreservation, patient‐reported outcomes, antitumour efficacy and safety. Fifty‐two patients received trilaciclib prior to E/P/A and 53 patients received placebo. Compared to placebo, administration of trilaciclib resulted in statistically significant decreases in the mean duration of SN in Cycle 1 (0 vs 4 days; P < .0001) and occurrence of SN (1.9% vs 49.1%; P < .0001), with additional improvements in red blood cell and platelet measures and health‐related quality of life (HRQoL). Trilaciclib was well tolerated, with fewer grade ≥3 adverse events compared with placebo, primarily due to less high‐grade haematological toxicity. Antitumour efficacy outcomes were comparable. Administration of trilaciclib vs placebo generated more newly expanded peripheral T‐cell clones (P = .019), with significantly greater expansion among patients with an antitumour response to E/P/A (P = .002). Compared with placebo, trilaciclib administered prior to E/P/A improved patients' experience of receiving treatment for ES‐SCLC, as shown by reduced myelosuppression, and improved HRQoL and safety profiles.

trilaciclib on additional measures of myelopreservation, patient-reported outcomes, antitumour efficacy and safety. Fifty-two patients received trilaciclib prior to E/P/A and 53 patients received placebo. Compared to placebo, administration of trilaciclib resulted in statistically significant decreases in the mean duration of SN in Cycle 1 (0 vs 4 days; P < .0001) and occurrence of SN (1.9% vs 49.1%; P < .0001), with additional improvements in red blood cell and platelet measures and health-related quality of life (HRQoL). Trilaciclib was well tolerated, with fewer grade ≥3 adverse events compared with placebo, primarily due to less high-grade haematological toxicity. Antitumour efficacy outcomes were comparable. Administration of trilaciclib vs placebo generated more newly expanded peripheral T-cell clones (P = .019), with significantly greater expansion among patients with an antitumour response to E/P/A (P = .002). Compared with placebo, trilaciclib administered prior to E/P/A improved patients' experience of receiving treatment for ES-SCLC, as shown by reduced myelosuppression, and improved HRQoL and safety profiles. and thrombocytopenia. [2][3][4][5][6][7] The adverse consequences of chemotherapy are particularly relevant in SCLC, where more than half of patients are aged ≥65 years at diagnosis and patients often present with multiple comorbidities. As such, patients with SCLC are more likely to experience clinically significant side effects related to CIM. 8,9 The haematological toxicities of CIM are a major source of morbidity, mortality and cost among patients with cancer, due to an increased risk of infection, sepsis, bleeding and fatigue. Currently, CIM is managed with dose delays and reductions that reduce the dose intensity and potentially, the antitumour efficacy, of chemotherapy, as well as intervention with growth factors or transfusions. 10 Consequently, preventing CIM is an important goal in the treatment of patients with cancer.
Chemotherapy remains a major component of treatment for patients with SCLC; however, despite high initial response rates, longterm outcomes remain poor. Cytotoxic drugs can cause tumour cells to become more susceptible to immune destruction by stimulating antigen presentation and T-cell priming. 11 However, tumour cells have developed multiple mechanisms for evading immune surveillance. Combining immune checkpoint inhibitors (ICIs) with chemotherapy may disrupt these escape mechanisms and efficiently restore the antitumour activity of the immune system. 12 Considering the success of ICI for the treatment of non-SCLC, 12 13 More recently, the PD-L1 antibody, durvalumab, was also approved by the FDA in combination with etoposide plus platinum chemotherapy as a first-line treatment for adult patients with ES-SCLC, based on the results of the Phase III CASPIAN trial. 14 Although the combination of chemotherapy plus ICIs may provide improved survival for some patients, further research is required to optimise treatment outcomes for patients with ES-SCLC.

Chemotherapy indiscriminately kills proliferating cells, including
HSPCs and immune cells, and therefore the full benefit of

What's new
• Trilaciclib is known to protect immune function and blood progenitor cells from chemotherapy-induced damage.
Might it improve outcomes for patients with small-cell lung cancer (SCLC)? In this prospective, randomized study, the authors found that when trilaciclib was given prior to treatment with chemotherapy plus atezolizumab, it reduced myelosuppression and the need for supportive care. It also enhanced T-cell immunity, and improved quality of life. Trilaciclib thus has considerable potential as a new standard of supportive care for SCLC patients receiving myelosuppressive chemotherapy. chemotherapy plus ICI combinations may not be realised due to resulting CIM and immunosuppression. 15 An intervention that both prevents CIM and maintains immune system function when cytotoxic treatment is administered could therefore reduce the adverse consequences of chemotherapy and potentially augment the antitumour efficacy of chemotherapy plus ICI combination regimens.
Trilaciclib is being developed as a first-in-class myelopreservation therapy to prevent CIM in adult patients with ES-SCLC. 16,17 Trilaciclib is an intravenous (IV) CDK4/6 inhibitor that transiently arrests CDK4/6-dependent cells, including HSPCs and lymphocytes, in the G1 phase of the cell cycle, thereby preventing them from proliferating in the presence of cytotoxic chemotherapy. In doing so, trilaciclib provides resistance to chemotherapy-induced damage, and favourably alters the tumour immune microenvironment through transient T-cell inhibition, with differential effects on T-cell subsets. [16][17][18] The myelopreservation benefits of trilaciclib have been evaluated in patients receiving chemotherapy for the treatment of SCLC because SCLC tumour cells replicate independently of CDK4/6 due to the obligate loss of retinoblastoma.
This allows the assessment of trilaciclib's effects on the host while minimising theoretical concerns related to effects on the tumour. 19 In addition to protecting lymphocyte populations and increasing immune activation through differential T-cell recovery, trilaciclib and other CDK4/6 inhibitors have been shown to enhance antitumour responses through other mechanisms in preclinical models, including enhanced T-cell activation through modulation of nuclear factor of activated T-cell activity and upregulation and stabilisation of PD-L1 expression on tumour cells, resulting in increased sensitivity to ICI. 18,20 Preclinically, the addition of trilaciclib to chemotherapy/ICI combinations has been shown to enhance and prolong the duration of the antitumour responses, providing a rationale for combining trilaciclib with chemotherapy/ICI regimens in patients with cancer. 18 Data from a randomised, double-blind, placebo-controlled Phase II trial of trilaciclib administered prior to E/P therapy in patients with newly diagnosed ES-SCLC showed myelopreservation benefits across multiple haematopoietic lineages (neutrophils and red blood cells [RBCs]) compared with placebo, with patients requiring fewer supportive care interventions and chemotherapy dose reductions. 21 The current study was designed to confirm the effects of trilaciclib on CIM in patients with newly diagnosed ES-SCLC treated with E/P and to investigate if the immune-enhancing effects of trilaciclib would translate to an improvement in the antitumour efficacy of atezolizumab.

| Study design and participants
This was a global, randomised, double-blind, placebo-controlled, multi-

| Randomisation and procedures
Patients were randomised 1:1 in a blinded manner to receive trilaciclib or placebo prior to etoposide, carboplatin and atezolizumab (E/P/A) therapy. An interactive web-response system was used to randomise patients according to a randomisation schedule generated by an unblinded statistician. Randomisation was stratified by ECOG PS (0/1 vs 2) and the presence of brain metastases (yes or no).
Patients were treated in an induction phase and a maintenance phase. During induction, patients received trilaciclib or placebo prior to

| Outcomes
The primary objective of the study was to evaluate the myelopreservation efficacy of trilaciclib vs placebo when administered prior to myelosuppressive chemotherapy plus atezolizumab. All myelopreservation analyses included laboratory value data rather than adverse events (AEs), unless otherwise specified. Primary endpoints were the duration of severe neutropenia (DSN) in C1 and percentage of patients with severe neutropenia (SN) (occurrence) during the treatment period. SN was defined as absolute neutrophil count (ANC) <0.5 × 10 9 cells per L. were evaluated as part of the safety assessments.

| PD-L1 immunohistochemistry
For the PD-L1 analysis, archival tissue was collected from each patient. The VENTANA PD-L1 (SP142) immunohistochemical assay (Roche) was used to assess expression of PD-L1, using the rabbit monoclonal anti-PD-L1 clone SP142. An IgG antibody was used as a negative control and human tonsil tissue as a positive control. Digital images of each section were generated using an Aperio Scanscope. Samples were considered negative or positive if <1% or ≥1% of the total tumour area (including stroma and inflammatory regions) contained PD-L1-labelled immune cells, respectively. All analyses were completed by Epistem, Ltd (Manchester, UK).

| Flow cytometry analysis
The ability of trilaciclib to preserve immune system function was assessed by measuring change from baseline in immune cell subsets.

| T-cell receptor β CDR3 analysis
To assess the effect of trilaciclib on the peripheral T-cell compartment and clonal expansion, T-cell receptor (TCR) β CDR3 regions were amplified and sequenced from purified genomic DNA in peripheral blood mononuclear cells isolated from whole blood samples at iC1D1 and mC1D1 using the immunoSEQ Assay (Adaptive Biotechnologies, Seattle, US) (Supplementary Methods). Newly detected expanded clones were defined as clones that were not detected at baseline, but were measurable at mC1D1.

| Statistical analysis
The planned sample size of the study was 106 (approximately 53 per group). Sample size was calculated to support the evaluation of trilaciclib prior to E/P/A vs placebo prior to E/P/A on each of the primary endpoints, with at least 90% power at a two-sided significance level of .025 (Bonferroni split of overall 2-sided α = .05 between the two primary endpoints). The assumed treatment effects on DSN in C1 and occurrence of SN were a between-group mean difference of 2 days (SD 2.5), and an absolute reduction of 34% (assuming a placebo event rate of 45%), respectively. The sample size was adjusted for the possibility that 5% of patients would not have any postbaseline ANC assessments.
The intention-to-treat (ITT) analysis set, used for myelopreservation, PROs and PFS/OS endpoints, included all randomised patients, with data analysed by randomly assigned treatment. Safety analyses included all randomised patients who received at least one dose of study drug, with data analysed by actual received treatment. Analyses of tumour response were performed in patients who had a measurable target lesion at baseline, and had either at least one postbaseline tumour assessment, discontinued treatment due to clinical progression or died due to disease progression before their first postbaseline tumour scan. There were two clinical database locks were used for safety and antitumour efficacy evaluation. DSN in C1 was evaluated using a nonparametric analysis of covariance. 27 Patients who did not have SN in C1 were assigned a value of 0. For binary endpoints (eg, occurrence of SN or RBC transfusion on or after Week 5), treatment effect was assessed using a modified Poisson regression model. 28 For counting endpoints, a negative binomial regression model was used to evaluate treatment effect. All statistical models included treatment, ECOG PS (0/1 vs 2) and brain metastases (yes or no) as fixed effects, with corresponding baseline value as a covariate. For the primary and key secondary efficacy endpoints, a Hochberg-based gatekeeping procedure 29 was utilised to allow strong control of family-wise Type I error rate at one-sided α = .025 level. Mean difference in DSN in C1, with one-sided, multiplicity-adjusted P values and 95% confidence intervals (CI) is reported. Adjusted relative risk (aRR) and 95% CI are reported for all other binary and counting endpoints.
A post hoc analysis of DSN in C1, occurrence of SN and occurrence of RBC transfusion on or after Week 5 was evaluated by age subgroup (<65 and ≥65 years). The same statistical models were applied to each group to estimate the treatment effect of trilaciclib vs placebo.
Tumour response status per RECIST v1.1 was derived from measurements provided by the investigator. ORR and its exact 95% CI using the Clopper-Pearson method were computed for each treatment group. The treatment effect was evaluated using a stratified Cochran-Mantel-Haenszel method. DOR was characterised using the Kaplan-Meier method for patients who achieved a complete or partial response. The Kaplan-Meier method was used to estimate median PFS and OS; treatment group difference was evaluated using a stratified log-rank test, with the hazard ratio (HR) and its 95% CI generated from a Cox proportional hazard model. OS data are considered mature when at least 70% of deaths have occurred (not reached at the time of the second DBL). Safety measures are summarised using descriptive statistics, except for hospitalisation due to CIM or sepsis, where treatment group differences were assessed using a modified Poisson model and incidence rates using a negative binomial model. All statistical analyses were conducted using SAS software, v.9.4.  Figure S1). Two patients were randomised to receive trilaciclib but did not receive any study drug (one patient did not meet eligibility criteria and was randomised in error and one patient's platelet count did not meet dosing criteria on C1D1). Baseline demographics and disease characteristics were similar between the treatment groups. Expression of PD-L1 was detected in 18/48 (37.5%) tumour tissue samples, including 8/21 (38.1%) in the trilaciclib group and 10/27 (37.0%) in the placebo group (Table 1).

| Myelopreservation
Trilaciclib administered prior to E/P/A therapy reduced chemotherapyinduced neutropenia compared with placebo, as measured by statistically significant improvements in the primary endpoints of DSN in C1 and occurrence of SN (Figure 1; Supplementary raw P value = .0343, multiplicity-adjusted P value = .0686) (

| Patient experience
PRO completion rates were high (>92% in both groups) throughout the study. At the end of C4, there were no differences between the treatment groups in the adjusted mean difference from baseline for total or any subscale score. A trend favouring trilaciclib was observed at the end of C2 for several domains. For FWB, the least squares

| Immunomodulatory effects
Flow cytometric analysis of T-cell populations showed that compared with placebo, patients receiving trilaciclib had a higher ratio of CD8+ T cells to Tregs and activated CD8+ T cells to Tregs, through PTV +90, although differences between the treatment groups were not statistically significant ( Figure 4A,B). Immunosequencing analyses showed that patients in the trilaciclib group had a significantly higher number of expanded T-cell clones at the end of induction compared with patients receiving placebo (P = .019; Figure 4C). Patients in the trilaciclib group with an antitumour response to E/P/A had significantly more clonal expansion than responders who received placebo (P = .002) and more than nonresponders who received trilaciclib (P = .016; Figure 4D). Responders receiving trilaciclib also had more newly detected expanded clones and a significant increase in the    Clinical concerns over toxicities associated with CIM often lead to dose reductions and/or delays, which are more frequent in elderly patients and can be associated with poorer therapeutic outcomes. 10 There are a number of potential reasons why the findings in the current study did not translate into significant improvements in antitumour efficacy. The effects of trilaciclib on antitumour efficacy are predicted to be primarily driven by the tumour type, chemotherapy type, and host. Specifically, the tumour type must be sufficiently responsive to chemotherapy such that maintenance of chemotherapy dose intensity is beneficial, the chemotherapy should promote immune activation, and the host must be able to mount an effective cytolytic response against the tumour. ES-SCLC is a highly aggressive tumour that typically recurs and progresses rapidly despite initial response to chemotherapy; among patients treated with chemotherapy and ICIs, median survival remains just 12 months from diagnosis. 13 Although SCLC has a high tumour mutation burden, it is not considered particularly immunogenic or sensitive to immune modulation. 12,52,53 PD-L1 expression has been reported to be low (<50%) in clinical studies, and in our study, was less than 40%. Moreover, SCLC tumours have reduced expression of major histocompatibility complex class I and class II molecules, a known immune escape mechanism, reflecting a less immunogenic environment. 53,54 Nevertheless, the median OS of 12 months with trilaciclib prior to E/P/A observed in our study is consistent with that seen with E/P/A in the pivotal IMPower133 study, 13 indicating that, while there was no improvement in the antitumour efficacy of chemotherapy plus atezolizumab, trilaciclib did not antagonise the effects of E/P/A in patients with ES-SCLC. This is consistent with previous results indicating that the myelopreservation benefits of trilaciclib are not accompanied by detrimental effects on the efficacy of standard-of-care chemotherapy regimens. 21,30 A limitation of our study is the small sample size, which may have reduced the ability to observe statistically significant treatment effects on secondary myelopreservation measures, such as occurrence of FN AEs, infections and antibiotics usage. However, large treatment effects in these endpoints were not expected, given that patients in both arms could receive supportive care interventions for CIM, apart from in C1 where the use of prophylactic G-CSF and ESAs was prohibited. As a result, the frequency of these secondary events was expected to be low, and the difference between treatment groups to be small. Furthermore, the small sample size meant that it would only be possible to detect large differences in OS. The observed immune effects of trilaciclib are therefore hypothesis generating and require further investigation.
Overall, the results of our study confirm the myelopreservation benefits of trilaciclib administered prior to chemotherapy for SCLC, as demonstrated by the reduction in clinically significant CIM and a reduction in the use of standard-of-care interventions. Consistent with these myelopreservation benefits, trilaciclib was associated with

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.

ETHICS STATEMENT
The study (NCT03041311) was designed and conducted in compli-