Research Report

The Impact of a Clinical Decision Support System for Advanced Gastric or Gastroesophageal Junction Cancer

Authored by

Scott Paulson, MD1; Lisa M Hess, PhD2; Anindya Chatterjee, PhD2; Marley Boyd, MS3; Kathleen M Aguilar, MPH3; Astra M Liepa, PharmD2

Affiliation

1Texas Oncology Baylor Charles A Sammons Cancer Center, Dallas, TX
2Eli Lilly and Company, Indianapolis, IN
3Ontada, The Woodlands, TX

Disclosures

Beyond employment, Dr Liepa reports stock ownership in Eli Lilly and Company. Dr Paulson reports advisory board payment from Incyte, AAA Pharmaceutical, Ipsen, QED, Bristol Meyers Squibb, Eisai, and Exelixis; research grants from AstraZeneca, Ipsen, and Hutchinson; stock ownership in Seagen; and research consulting services payment from AstraZeneca and Eli Lilly and Company. Mr Boyd and Ms Aguilar report research consulting services payment from Eli Lilly and Company. Drs Chatterjee and Hess report no relevant financial relationships beyond employment.

Author Contributions: Drs Paulson, Hess and Liepa report that they contributed to the conception, design, and interpretation of the study, as well as critical review of the manuscript. Mr Boyd reports that he contributed to acquisition of the data, analysis, and interpretation of the study, as well as critical review of the manuscript. Dr Chatterjee reports that he contributed to the analysis and interpretation of the work, as well as drafting and critical review of the manuscript. Ms Aguilar reports that she contributed to conception, design, and interpretation of the work, as well as drafting and critical review of the manuscript. All authors read and approved the final manuscript.

Citation

J Clin Pathways. 2020;6(10):55-64, s1-s6. doi:10.25270/jcp.2020.12.00001
Received October 23, 2020; accepted November 19, 2020.

Acknowledgements

The authors would like to acknowledge Lisa Kaspin-Powell (McKesson Life Sciences) for medical writing assistance with the preparation of this manuscript.

Correspondence

Kathleen M Aguilar
Ontada
10101 Woodloch Forest Dr.
The Woodlands, TX 77380
Phone: (562) 212-9250
Email: Kat.aguilar@ontada.com

Abstract: Gastric cancer treatment heterogeneity was assessed over successive updates of clinical pathways. Methods: Patient profiles, treatment patterns, and outcomes were assessed descriptively among adult patients with advanced gastric or gastroesophageal junction cancer who initiated treatment September 1, 2009 through March 31, 2018. Concordance with National Comprehensive Cancer Network recommendations was assessed based on the documented regimens in the electronic health record with guidelines in effect during treatment. The Herfindahl-Hirschman Index (HHI) was used to evaluate treatment heterogeneity. Results: 2297 patients were included in the analysis. Before Pathways implementation, 25.2% of patients received first-line, guideline-concordant regimens; for the last evaluation period, 66.0% of patients received concordant regimens. HHI values were <0.1 (very heterogeneous) before Pathways implementation and >0.2 (more homogenous) during the last evaluation period. Conclusion: Higher concordance with guidelines and lower heterogeneity over the study period were observed. Further investigations are needed to understand relationships between clinical outcomes, treatment heterogeneity, and guideline concordance.

Key Words: stomach neoplasms, retrospective studies, decision support systems, treatment outcomes, clinical pathways


Previous research has demonstrated that treatment of gastric or gastroesophageal junction (GEJ) cancer can be highly variable, with significant heterogeneity observed in treatment choice for this population.1-5 Treatment selection is influenced by patient- and disease-related factors as well as differences across practice sites.1,5 Inconsistent treatment patterns may contribute to poor clinical outcomes, reduced patient quality of life, and health care resource use variation.1,6,7 Therefore, it is hypothesized that gastric/GEJ cancer treatment regimens concordant with clinical guidelines will reduce heterogeneity and improve clinical outcomes.

The US Oncology Network (The Network) is a community-based network of over 40 oncology practices with 470 sites of care that treat over 1 million patients annually.8 Care based on National Comprehensive Cancer Network (NCCN) Guidelines® is promoted within The Network through Value Pathways powered by NCCN™ (Pathways), that are integrated into the electronic health record (EHR) of The Network, iKnowMedSM (iKM), which is used by all Network clinics evaluated in this review.9,10 Development of Pathways is physician-led in collaboration with the NCCN panel members. Pathways are not guided by payer mandates nor are providers reimbursed by payers based on adherence to Pathways. At the point of care, Pathways provide evidence-based treatment options specific for individual patients with consideration of the predicted efficacy, toxicity, and associated costs. Previous research of patients with non-small cell lung or colon cancer reported that adherence to Pathways was associated with lower treatment costs and comparable clinical outcomes.9,11,12

In September 2010, The Network implemented the first iteration of the Level I Pathway for gastric cancer, which was integrated into iKM, for specific lines of therapy.13 In December 2014, a decision support tool, Clear Value PlusSM was released and included a new version of Pathways, Value Pathways powered by NCCN for gastric cancer. Clear Value Plus allows physicians to access evidence-based clinical content at the point of care, along with recommendations aligned with NCCN Guidelines.14 The gastric cancer Pathway is updated when new evidence presents itself; two significant reviews and updates took place in March 2017 and May 2018.

The aim of this study was to descriptively assess treatment heterogeneity across these successive time points—starting with 2010 and at each subsequent Pathway update in 2014, 2017, and 2018—among patients with advanced gastric/GEJ cancer treated in the community oncology setting.

Methods

Study Design and Data Sources

This study was reviewed and granted an exception and waiver of consent by The US Oncology Inc. Institutional Review Board. This was a retrospective observational study of adult patients with advanced gastric/GEJ cancer who initiated first-line (1L) treatment within The Network between September 1, 2009 and March 31, 2018. Study data were obtained from structured fields of iKM, with supplementary vitality information provided by the Social Security Administration’s Death Master File.

Patients were followed through March 31, 2019; date of death; or last visit date. Patients were excluded if they participated in a clinical trial, had another primary cancer diagnosis, had fewer than two Network visits, received care at a Network site without full iKM EHR capacities, had data that was inaccessible for research, or had a treatment duration less than one day.

Statistical Analyses

This was a hypothesis-generating study, and all analyses were descriptive in nature. Patient demographics, clinical and treatment characteristics were presented as categorical or continuous values. Concordance with NCCN treatment recommendations was assessed by Pathways based on NCCN Guidelines that were in effect during Pathway period. Note, for description of treatment regimens and calculations of treatment variability, fluorouracil and capecitabine were considered therapeutically interchangeable and were merged into a single “fluoropyrimidines” category. Likewise, cisplatin and carboplatin were merged into a single “platinum” category.

Treatment variability was measured by the Herfindahl-Hirschman Index (HHI).1,6,15 The HHI utilizes the sum of the squared proportion of patients using each regimen by line of therapy. HHI values range from 0 to 1, with lower values indicating more heterogeneity and higher values indicating more homogeneity. n optimum level of treatment heterogeneity for health care has not been established. Very low HHI values suggest high variability and inconsistent treatment, while very high HHI values suggest lack of patient-tailored treatment. For this study, therefore, a priori thresholds were established with HHI values <0.1 being considered very heterogeneous and values >0.2 more homogeneous.

Kaplan-Meier methods were used to assess time to treatment discontinuation (TTD) and overall survival (OS). Patients without a date of discontinuation were censored at the date of death, last visit date, or end of the study observation period, whichever occurred first. OS was measured from 1L initiation until death. Patients without a death date were censored at the last visit date or end of the study observation period, whichever occurred first. Time-to-event analyses were conducted and endpoints were compared using log-rank tests between each of the pre- and post-Pathway cohorts. No adjusted analyses were performed.

An alpha level of 0.05 was the primary criterion for statistical significance. Analyses were conducted using SAS® 9.4 (SAS Institute Inc, Cary, NC, US).

Results

From the eligible study population, 8 cohorts of patients were identified who initiated treatment before and after each of the four Pathway update periods (ie, September 2010, December 2014, March 2017, and May 2018). These cohorts, along with the Pathway update timeline, are depicted in Figure 1. Note, the 8 cohorts considered for this analysis are not mutually exclusive, as the “pre” cohorts in each pair represent the cumulative counts of patients who meet study criteria from September 1, 2009 until the start of each “post” cohort. Figure 1

Demographic and Clinical Characteristics

In total, 2297 patients with advanced gastric/GEJ cancer met eligibility criteria and were included in the analysis (Figure 1). From initiation of 1L treatment, the mean follow-up duration was 12.3 months (SD 15.9; Table 1). The longest mean follow-up duration of 18.1 months (SD 25.5) was observed among patients who initiated treatment prior to September 2010, while the shortest mean follow-up duration of 6.7 months (SD 4.3) was observed among those who initiated treatment after May 2018. The median age of the population was 65.3 years (range 20.6, 90+), with 59.6% male and 60.6% white. While stage at diagnosis was unavailable for 22.0% of the study population, most were diagnosed with stage III (10.6%) or stage IV (54.4%) disease. Over half (51.4%) were covered by Table 1 (continued)Medicare and 64.1% received care in Southern regions of the United States. Table 1

Treatment Patterns

Across the study observation period, 91 unique 1L regimens were observed. Prior to September 2010, 15.8% of patients received 1L fluoropyrimidines, 14.6% a platinum-based triplet regimen with docetaxel and a fluoropyrimidine, and 8.9% received fluoropyrimidine with oxaliplatin (Figure 2). The proportion of patients who received 1L fluoropyrimidine with oxaliplatin was 49.1% after May 2018. Regimens received by less than 3% of the total study population were classified into an “other” category. In each of the Pathway periods, approximately 30% of the cohort received treatments classified as “other” (range: 25.0% after December 2014, 34.9% after September 2010). Figure 2

Across the study population, 41.0% and 23.7% of patients received second-line (2L) and third-line (3L) treatment, respectively, during the study observation period (Table 1). In total, 85 unique 2L regimens were observed. Prior to September 2010, the most common 2L regimen was platinum plus paclitaxel (13.1%), and 63.9% of the study population received a 2L regimen classified into an “other” category (Figure 3). The proportion of patients who received an “other” regimen was 13.2% after May 2018. Figure 3

Prior to September 2010, 25.2% of 1L regimens documented in the EHR were concordant with NCCN Guidelines (Figure 4). After May 2018, 66.0% of patients in the 1L setting and 66.7% in the 2L setting had documentation in the EHR of regimens concordant with NCCN Guidelines. Note, prior to September 2010, NCCN Clinical Practice Guidelines did not provide treatment recommendations for 2L or subsequent treatment, which is why this concordance was not presented. Figure 4

In the 1L setting, the HHI value (measure of heterogeneity) was 0.0853 (very heterogeneous) prior to September 2010 and 0.2568 (more homogenous) after May 2018
(Figure 5). In the 2L setting, the HHI value was 0.0555 (very heterogeneous) prior to September 2010 and 0.2952 (more homogenous) after May 2018. Figure 5

Clinical Outcomes

In the 1L setting, significant differences in TTD were observed between the pre- and post-periods of the March 2017 and May 2018 updates (log-rank P value < .01 for both; Table 2 and Supplemental Figures 1a-1d). Before and after the March 2017 update, median 1L TTD were 2.1 months (95% CI, 2.1, 2.3) and 2.8 months (95% CI, 2.3, 3.3), respectively. Before and after the May 2018 update, median 1L TTD were 2.3 months (95% CI, 2.1, 2.4) and 3.3 months (95% CI, 2.1, 5.1), respectively. Table 2

Significant differences in 2L TTD were also observed between the pre- and post-periods of the March 2017 and May 2018 updates (log-rank P value = .02 for both; Table 2 and Supplemental Figures 2a-2d). Median 2L TTD were 1.5 months (95% CI, 1.4, 1.8) and 1.9 months (95% CI, 1.4, 2.6) before and after the March 2017 update, respectively. Median TTD in the 2L setting was 1.6 months (95% CI 1.4, 1.9) among patients who received treatment prior to May 2018 and 2.8 months (95% CI, 2.0, 7.9) among those who received treatment after May 2018.

The high level of censoring among patients treated in the “post” periods following updates limited the ability to detect statistical differences in OS between any of the pre-post comparisons (Table 2). The median OS among patients who received treatment prior to May 2018 was 12.6 months (95% CI, 11.6, 13.3), while the median OS was not reached among those who received treatment after May 2018 (95% CI, 10.6, not reached; log-rank P=.15; Table 2 and Supplemental Figures 3a-3d).

Discussion

The results of this study suggest that treatment heterogeneity was reduced in both the 1L and 2L settings for patients with gastric/GEJ cancer as reported in the EHR during successive Pathway updates. This may be due to emerging evidence on therapeutic options that evolved the treatment landscape. Likewise, with each update of Pathways, if the number of regimens included in the Pathways recommendations decreased, it would be expected that treatment heterogeneity would decrease as well. Ultimately, the trends observed in the treatment landscape reflect the intent of the Clear Value Plus decision support system, which is to provide physicians with resources at the point-of-care to determine the most appropriate evidence-based standard of care for their patients in hopes of improving clinical outcomes.16 Of note, providers do not receive incentives from payers for adherence to preferred Pathway regimens, although individual clinics within The Network may offer incentives.

In the 1L setting, 8.9% of patients received fluoropyrimidine with oxaliplatin prior to September 2010 and 49.1% received this regimen after May 2018. Currently, fluoropyrimidine with oxaliplatin is a preferred 1L regimen (NCCN Guidelines Version 3.2020), given the results of trials that demonstrated favorable clinical outcomes and toxicity profiles associated with this regimen.17-19 Publication of these results likely influenced the increased adoption of fluoropyrimidines with oxaliplatin across The Network as the predominant 1L regimen among patients with gastric/GEJ cancer.

Compared with 1L treatment, 2L and subsequent treatment of gastric and GEJ cancers is less established. In 2009, at the start of this study’s observation period, there were no recognized standards of care for 2L treatment, and it was uncertain whether additional treatment was preferable to best supportive care alone.20 In 2011, Thuss-Patience et al reported that irinotecan monotherapy significantly improved survival compared to best supportive care among patients with previously treated metastatic or locally advanced gastric/GEJ cancer enrolled in the trial.21 Subsequently, several other trials have affirmed the benefit of 2L and subsequent treatment with chemotherapy or biologic therapies, including ramucirumab and pembrolizumab.22-25

Likewise, more use of pembrolizumab in the 2L setting was observed after March 2017. This trend may reflect an evolving treatment landscape that was influenced by publication of trials that demonstrated clinical benefits and a tolerable safety profile associated with use of pembrolizumab among previously treated patients with gastric/GEJ cancer.23-25

While outcomes were descriptive in nature and were not adjusted for baseline covariates, the median TTD was approximately 1 month longer among patients treated after May 2018 compared with those treated prior to May 2018 in both the 1L and 2L settings (P<.01 and P=.02, respectively). This finding can be considered hypothesis-generating and should be further evaluated in future research. While reasons for treatment discontinuation and use of supportive care were not assessed for this study, prior research has demonstrated that patients with advanced gastric/GEJ cancer often discontinue treatment due to decreased performance status, progression, and/or toxicity.26 Current NCCN Guidelines recommend 1L and subsequent treatment choices be based on expected efficacy and toxicity, with preference given to regimens with more tolerable safety profiles.17

The increased availability of biologic therapies for the treatment of advanced gastric/GEJ cancer may improve survival durations and future, longitudinal research should be performed to confirm these trends. Wagner et al performed a Cochrane meta-analysis of 60 randomized controlled trials and found that the median OS of patients with gastric and GEJ cancer treated with chemotherapy was approximately 11 months.27 In this study, median OS ranged from 12.1 months (after September 2010) to not reached (after May 2018) among patients treated with chemotherapies and biologic therapies. Comparisons between these real-world results and clinical trials are limited by differences in patient population characteristics, as well as methods of follow-up assessment. With the availability of 2L treatment options, OS may be extended in more contemporary studies vs clinical trials that predated these advancements. Due to variable duration of follow-up, assessment of OS was limited for those patients who initiated care during later periods of the study, as these patients had less follow-up time than those who initiated earlier, limiting the ability to make comparisons across periods.

Data from this study were sourced from a network of community-based oncology practices that had implemented a decision support system for treatment of gastric/GEJ cancer, which offered a unique opportunity to evaluate treatment patterns with a large patient sample. By sourcing data solely from Network clinics, however, the generalizability of this study may be limited, as there may be differences in the practice patterns and patient profiles compared to other community oncology clinics.

Data were limited by the lack of information about health care utilization outside The Network, including cost of care, hospitalizations, emergency department visits, and treatment from other oncology clinics. If patients received any treatment outside of The Network, some treatment sequences may have been misrepresented. For example, patients who appear to have initiated 1L treatment may have had prior 1L treatment outside The Network. Additionally, patients without evidence of 2L or 3L treatment may have had ongoing 1L therapy, received their next treatment outside The Network, entered hospice care, or died.

The structured EHR data used for this study were generated during routine patient care and not collected for research purposes, thus, data entry errors and incomplete records could not be verified nor corrected. For example, some variables of interest were incomplete across the study population, including tumor location, ECOG performance status, histology, and human epidermal growth factor receptor 2 (HER2) status. Despite this variation, it is unlikely that the measurement of treatment variability (HHI score) was influenced, as its performance is unaffected by low sample sizes and is generally used with as few as 20 patients without issue.1 There was also variation in the degree of missing information across Pathways periods studied.

This study was descriptive in nature and did not account for other changes to the EHR and practice patterns that may have influenced these results. For example, payer-led performance initiatives have been introduced and are often guided by Value Pathways powered by NCCN and NCCN Guideline recommendations. Despite the fact that Pathways evaluated in this study are not guided by payer mandates nor are providers reimbursed by payers based on adherence to Pathways, future investigations may consider how participation in these programs influences treatment decisions and, in turn, treatment heterogeneity. Additionally, outcome assessments were not adjusted to control for underlying baseline demographic or clinical differences between the groups. Therefore, causality cannot be inferred as due to Pathways alone.

Conclusion

This study observed reduced treatment heterogeneity and improved utilization of regimens in concordance with NCCN Guidelines as documented in the EHR across successive updates of Pathways for gastric cancer, which was deployed within a large network of community oncology practices. The modest increase observed for median TTD over time should be explored further to determine if this improvement is statistically associated with reduced treatment heterogeneity and/or EHR-reported concordance with NCCN Guidelines.   

Supplementary materials can be viewed in the PDF.

References

1. Abrams T, Hess LM, Zhu YE, Schelman W, Liepa AM, Fuchs C. Predictors of heterogeneity in the first-line treatment of patients with advanced/metastatic gastric cancer in the U.S. Gastric Cancer. 2018;21(5):738-744. doi:10.1007/s10120-018-0802-5

2. Snyder RA, Penson DF, Ni S, Koyama T, Merchant NB. Trends in the use of evidence-based therapy for resectable gastric cancer. J Surg Oncol. 2014;110(3):285-290. doi:10.1002/jso.23635

3. Dudeja V, Habermann EB, Zhong W, et al. Guideline recommended gastric cancer care in the elderly: insights into the applicability of cancer trials to real world. Ann Surg Oncol. 2011;18(1):26-33. doi:10.1245/s10434-010-1215-9

4. Zhao B, Blair SL, Katz MHG, Lowy AM, Kelly KJ. Adherence with operative standards in the treatment of gastric cancer in the United States. Gastric Cancer. 2019;23(3):550-560. doi:10.1007/s10120-019-01028-5

5. Mokdad AA, Ali A, Yopp AC, et al. Adoption of evidence-based novel therapies in the treatment of gastric cancer: A national observational study. Cancer. 2018;124(6):1122-1131. doi:10.1002/cncr.31179

6. Hess LM, Michael D, Mytelka DS, Beyrer J, Liepa AM, Nicol S. Chemotherapy treatment patterns, costs, and outcomes of patients with gastric cancer in the United States: a retrospective analysis of electronic medical record (EMR) and administrative claims data. Gastric Cancer. 2016;19(2):607-615. doi:10.1007/s10120-015-0486-z

7. Karve S, Lorenzo M, Liepa AM, Hess LM, Kaye JA, Calingaert B. Treatment patterns, costs, and survival among Medicare-enrolled elderly patients diagnosed with advanced stage gastric cancer: analysis of a linked population-based cancer registry and administrative claims database. J Gastric Cancer. 2015;15(2):87-104. doi:10.5230/jgc.2015.15.2.87

8. The US Oncology Network. Our Company. 2020. Accessed November 23, 2020. https://www.usoncology.com/our-company

9. Wilfong L, Ginsburg A, Hoverman JR, et al. Treatment pathways for first-line metastatic non-small cell lung cancer: cost and survival. J Clin Pathways. 2020;6(8):52-57. doi:10.25270/jcp.2020.10.00001

10. McKesson. Value Pathways powered by NCCN—Oncology Clinical Pathways 2019. Accessed November 23, 2020. https://www.mckesson.com/specialty/oncology-pathways/

11. Hoverman JR, Cartwright TH, Patt DA, et al. Pathways, outcomes, and costs in colon cancer: retrospective evaluations in two distinct databases. J Oncol Pract. 2011;7(suppl 3):s52-s59. doi:10.1200/JOP.2011.000318

12. Neubauer MA, Hoverman JR, Kolodziej M, et al. Cost effectiveness of evidence-based treatment guidelines for the treatment of non-small-cell lung cancer in the community setting. J Oncol Pract. 2010;6(1):12-18. doi:10.1200/JOP.091058

13. Neubauer M. Clinical pathways: reducing costs and improving quality across a network. Am J Manag Care. 2020;26(spec 2):sp60-sp61. doi:10.37765/ajmc.2020.42550

14. NCCN, McKesson Specialty Health, and The US Oncology Network Announce Plans for Integration with Elekta and Varian. News release. Business Wire; September 24, 2013. Accessed November 23, 2020. https://www.businesswire.com/news/home/20130924006632/en/The-National-Comprehensive-Cancer-Network-McKesson-Specialty-Health-and-The-US-Oncology-Network-Announce-Plans-for-Integration-with-Elekta-and-Varian

15. Weinstock DS. Using the Herfindahl Index to measure concentration. Antitrust Bulletin. 1982;27:285. Accessed November 23, 2020. https://heinonline.org/HOL/LandingPage?handle=hein.journals/antibull27&div=13&id=&page=

16. The National Comprehensive Cancer Network, The US Oncology Network and McKesson Specialty Health collaborate to offer new Value Pathways for cancer care. News release. McKesson; November 29, 2021. Accessed November 23, 2020. https://www.mckesson.com/About-McKesson/Newsroom/Press-Releases/2012/The-NCCN--The-US-Oncology-Network-and-McKesson-Specialty-Health-Collaborate-to-Offer-New-Value-Pathways-for-Cancer-Care/

17. National Comprehensive Cancer Network. Clinical Practice Guidelines in Oncology: Gastric Cancer Version 3.2020. August 14, 2020. Accessed October 13, 2020. https://www.nccn.org/professionals/physician_gls/default.aspx#gastric

18. Enzinger PC, Burtness BA, Niedzwiecki D, et al. CALGB 80403 (Alliance)/E1206: a randomized phase II study of three chemotherapy regimens plus cetuximab in metastatic esophageal and gastroesophageal junction cancers. J Clin Oncol. 2016;34(23):2736-2742. doi:10.1200/JCO.2015.65.5092

19. Al-Batran SE, Hartmann JT, Probst S, et al. Phase III trial in metastatic gastroesophageal adenocarcinoma with fluorouracil, leucovorin plus either oxaliplatin or cisplatin: a study of the Arbeitsgemeinschaft Internistische Onkologie. J Clin Oncol. 2008;26(9):1435-1442. doi:10.1200/JCO.2007.13.9378

20. Wesolowski R, Lee C, Kim R. Is there a role for second-line chemotherapy in advanced gastric cancer? Lancet Oncol. 2009;10(9):903-912. doi:10.1016/S1470-2045(09)70136-6

21. Thuss-Patience PC, Kretzschmar A, Bichev D, et al. Survival advantage for irinotecan versus best supportive care as second-line chemotherapy in gastric cancer--a randomised phase III study of the Arbeitsgemeinschaft Internistische Onkologie (AIO). Eur J Cancer. 2011;47(15):2306-2314. doi:10.1016/j.ejca.2011.06.002

22. Kanagavel D, Fedyanin M, Tryakin A, Tjulandin S. Second-line treatment of metastatic gastric cancer: Current options and future directions. World J Gastroenterol. 2015;21(41):11621-11635. doi:10.3748/wjg.v21.i41.11621

23. Muro K, Chung HC, Shankaran V, et al. Pembrolizumab for patients with PD-L1-positive advanced gastric cancer (KEYNOTE-012): a multicentre, open-label, phase 1b trial. Lancet Oncol. 2016;17(6):717-726. doi:10.1016/S1470-2045(16)00175-3

24. Fuchs CS, Doi T, Jang RW, et al. Safety and efficacy of pembrolizumab monotherapy in patients with previously treated advanced gastric and gastroesophageal junction cancer: phase 2 clinical KEYNOTE-059 trial. JAMA Oncol. 2018;4(5):e180013. doi:10.1001/jamaoncol.2018.0013

25. Shitara K, Ozguroglu M, Bang YJ, et al. Pembrolizumab versus paclitaxel for previously treated, advanced gastric or gastro-oesophageal junction cancer (KEYNOTE-061): a randomised, open-label, controlled, phase 3 trial. Lancet. 2018;392(10142):123-133. doi:10.1016/S0140-6736(18)31257-1

26. Kimura M, Kawachi S, Go M, et al. Effect of the timing of discontinuation of last-line chemotherapy on patient prognosis in advanced and recurrent gastric cancer. Mol Clin Oncol. 2019;10(1):173-179. doi:10.3892/mco.2018.1753

27. Wagner AD, Syn NL, Moehler M, et al. Chemotherapy for advanced gastric cancer. Cochrane Database Syst Rev. 2017;8(8):CD004064. doi:10.1002/14651858.CD004064.pub4