The incorporation of new therapies into routine cancer treatment requires regulatory review to ensure safety and efficacy, as well as a health technology assessment. This process is fraught with hurdles that can delay or hinder patient access.1 Health technology assessments aim to define the incremental benefits and costs of a new treatment compared to the current standard of care (or comparator). Randomized trials are the most robust way to demonstrate these differences. However, many recent drug approvals in lung cancer have been based on single arm studies in small, molecularly defined populations.
When novel agents demonstrate high rates of durable treatment response, comparative outcome data with standard treatments such as chemo- and immunotherapy as populations studied are often small. There is a clear opportunity for real-world evidence to supplement this gap, but regulators and payors have been inconsistent in their use of real-world evidence to facilitate regulatory and funding decisions.2 Additionally, once a product is approved and available in major markets such as the US, it is challenging to encourage drug manufacturers to mount additional clinical trials to enable approval and access in smaller markets. The end result leaves many patients without access to these novel, expensive therapies.
To address the lack of high quality comparative data, Ramagopalan et al. present a novel approach that could be applied when “sample size is insufficient, follow-up is immature, and overall survival estimates for the target population cannot be ascertained.”3 This is relevant to many novel agents in lung cancer, including TRK, ROS1, RET, MET, and RAF/MEK inhibitors among others. The authors conducted a “transportability” analysis comparing real-world outcomes of patients with lung cancer in the US and Canada.3
The authors undertook a pooled regression analysis, comparing survival outcomes for 15,669 patients in the US Flatiron database and 1,763 Canadian patients from the Alberta Cancer Registry. Patients were diagnosed with advanced lung cancer (stage IIIB/IV) between January 2011 and September 2020. They focused on two subgroups—patients that received initial platinum doublet chemotherapy and those that received initial pembrolizumab immunotherapy. They developed a complete-case fitted model to estimate survival rates in Canadian lung cancer patients based on data from the US cohort. Despite controlling for multiple covariates including age, sex, performance status, histology, and smoking status, there were differences between the cohorts. Patients in the Alberta cohort had more advanced disease (stage IV 82% vs. 68%; more metastatic sites 41% vs. 14%), more comorbidities (43% vs 27%) and a longer time from diagnosis to treatment (median 1.8 vs. 1.1 months) compared to the US cohort. Patients in the US cohort were more likely to have squamous histology (39% vs. 17%). After adjusting for these differences, the estimated survival rates from the US population more closely mirrored the Canadian reality.
The transported estimates for the first-line chemotherapy-treated cohort were similar to actual Canadian survival data, with an absolute mean difference of 0.56% over 5 years. The modelled survival estimates for US patients receiving first-line immunotherapy were higher than Canadian patient data, up to 4.54% higher over 2.5 years with the greatest overestimate in the first 9 months. Thus, use of the transported model would either approximate or underestimate the benefit of a novel therapy in the Canadian setting.
The use of datasets such as the Flatiron Clinicogenomic Database may help demonstrate the similarity or differences in outcomes of targeted therapy with standard chemo- or immunotherapy between these smaller oncogene driven subgroups and the overall NSCLC patient population. Regulators may argue about the relevance of this model based on unselected patients to outcomes in smaller molecularly defined populations.4 Are there ways to further strengthen this approach? At a technical level, greater understanding of the model framework, variables chosen, interaction terms, and handling of missing data would also contribute.
For example, single imputation was performed to account for missing performance status and smoking status data in the Canada cohort, while multiple imputation was used in the US cohort. Further understanding of the number of Canadian patients with missing data, the impact on model assumptions and the reasons for the different approaches would be helpful.
In addition, we must ask whether this can be confidently extrapolated beyond the province of Alberta to the rest of Canada and to other countries. Population differences exist between Canadian provinces including in race and ethnicity, tobacco use, frequency of NSCLC genomic alterations, survival outcomes, drug funding, and availability.5,6,7 These may all impact generalizability of the model. Ideally, we would have access to high caliber data integrated into large networks, to allow for modelling of data with minimal need for adjustment to address biases.8 Networks like this have been established in several countries including Canada, France, and Germany among others in addition to established databases in the US.
However, reliance on real-world evidence should not take the place of increasing international participation in clinical trials. Canada, for example, has made strong contributions to the global arena of cancer research, however opportunities for growth remain.9,10 We must preserve and grow enthusiasm, infrastructure, and funding for clinical trials and ensure we do not become reliant on data exclusively from other countries. The Canadian Cancer Trials Group’s Committee on Economic Analysis is focused on generating high quality economic data from cooperative group trials with Canadian patients.11
However, even with these efforts, not all practice-changing trials will include patients from every country. Thus, it is important to have methods like the proposed transportability analysis to translate the benefit of new treatments to different settings. It is also important to continue to support the development and maintenance of searchable databases of high-quality real-world evidence, such as the Flatiron and Alberta provincial databases used in this study.
Precision medicine poses unique challenges in which smaller subgroups of patients may require different approaches to establishing the efficacy, safety, and cost effectiveness of new personalized treatments.11 We believe that the value of and need for real-world evidence will increase over time, and extend to early stage as well as advanced lung cancer. It is our hope that the work of Ramagopolan et al.3 will help inform and expedite regulatory and funding approval of novel drugs in lung cancer when randomized data are not available, to promote faster and greater access to revolutionary new medicines for patients.
References
- 1. P. Wheatley-Price, “Access to Cancer Drugs in Canada,” Current Oncology 2022, Vol. 29, Pages 7583-7586, vol. 29, no. 10, pp. 7583–7586, Oct. 2022, doi: 10.3390/CURRONCOL29100598.
- 2. Griesinger F, Cox O, Sammon C, Ramagopalan SV, Popat S. Health technology assessments and real-world evidence: tell us what you want, what you really, really want. J Comp Eff Res. 2022 Apr;11(5):297-299. doi: 10.2217/cer-2021-0296. Epub 2022 Feb 25. PMID: 35212231.
- 3. S. V. Ramagopalan et al., “Transportability of Overall Survival Estimates From US to Canadian Patients With Advanced Non–Small Cell Lung Cancer With Implications for Regulatory and Health Technology Assessment,” JAMA Netw Open, vol. 5, no. 11, pp. e2239874–e2239874, Nov. 2022, doi: 10.1001/JAMANETWORKOPEN.2022.39874.
- 4. Popat S, Liu SV, Scheuer N, Hsu GG, Lockhart A, Ramagopalan SV, Griesinger F, Subbiah V. Addressing challenges with real-world synthetic control arms to demonstrate the comparative effectiveness of Pralsetinib in non-small cell lung cancer. Nat Commun. 2022 Jun 17;13(1):3500. doi: 10.1038/s41467-022-30908-1. PMID: 35715405; PMCID: PMC9205915.
- 5. “Profile table, Census Profile, 2021 Census of Population – Alberta [Province];Canada [Country].” https://www12.statcan.gc.ca/census-recensement/2021/dp-pd/prof/details/page.cfm?Lang=E&SearchText=canada&GENDERlist=1,2,3&STATISTIClist=1&DGUIDlist=2021A000248,2021A000011124&HEADERlist=0 (accessed Mar. 10, 2023).
- https://www.census.gov/quickfacts/fact/table/US/RHI125221 (accessed Mar. 10, 2023).
- 6. E. D. Nishri, A. J. Sheppard, D. R. Withrow, and L. D. Marrett, “Cancer survival among First Nations people of Ontario, Canada (1968-2007),” Int J Cancer, vol. 136, no. 3, pp. 639–645, Feb. 2015, doi: 10.1002/IJC.29024.
- 7. Lung Cancer Canada. Faces of Lung Cancer Report, November 2022. https://www.lungcancercanada.ca/LungCancerCanada/media/Images/Content/LCC1001ENG_FofLC_2022_Interior_V3.pdf (accessed Nov. 17, 2023)
- 8. Institute of Medicine (US), “Standardization to Enhance Data Sharing,” Mar. 2013, Accessed: Mar. 18, 2023. [Online]. Available: https://pubmed.ncbi.nlm.nih.gov/23700647/
- 9. S. Sharma et al., “Cancer, clinical trials, and Canada: Our contribution to worldwide randomized controlled trials,” Current Oncology, vol. 28, no. 2, pp. 1518–1527, Apr. 2021, doi: 10.3390/CURRONCOL28020143/S1.
- 10. R. Xu et al., “Advancing Academic Cancer Clinical Trials Recruitment in Canada,” Current Oncology, vol. 28, no. 4, pp. 2830 – 2839, Jul. 2021, doi: 10.3390/CURRONCOL28040248.
- 11. K. K. W. Chan et al., “The Past, Present, and Future of Economic Evaluations of Precision Medicine at the Committee for Economic Analyses of the Canadian Cancer Trials Group,” Current Oncology 2021, Vol. 28, Pages 3649-3658, vol. 28, no. 5, pp. 3649–3658, Sep. 2021, doi: 10.3390/CURRONCOL28050311.
