Patterns and Predictors of Treatment Choice in a Real-World Cohort of Patients with Metastatic Gastric Cancer in the United States

J Clin Pathways. 2021;7(7):34-39. doi: 10.25270/jcp.2021.09.3
Received April 7, 2021; Accepted July 19, 2021.

There are about 27,600 new cases of gastric cancer diagnosed in the United States (US) annually. Of these patients, 28% are diagnosed with locally-advanced disease and 36% have distant metastases at the time they are initially diagnosed.¹ Unfortunately, prognosis remains poor; only 5.2% of patients diagnosed with metastatic gastric cancer (mGC) are alive five years after their diagnosis. 

Treatment guidelines are available to support the evidence-based treatment of patients diagnosed with gastric cancer to ensure patients receive the best care and may achieve the best possible outcomes from this disease. The National Comprehensive Cancer Network (NCCN) publishes gastric cancer treatment guidelines for the US. These guidelines are updated regularly for ready identification of evidence-based treatments for cancer patients. The treatment of unresectable, advanced, or metastatic gastric cancer is supported by Category l evidence (ie, based on high-level scientific evidence and has consensus from NCCN gastric cancer panel members) in both the first- and second-line settings since 2010. Since thattime, despite a number of updates and changes in preferred treatments, similar agents and combinations have been included for the care of patients with gastric cancer; for example, docetaxel plus cisplatin plus fluorouracil, epirubicin plus cisplatin plus fluorouracil, and trastuzumab are among those treatments that have been included in treatment guidelines since 2010. Newer agents such as ramucirumab and immune checkpoint inhibitors (ie, nivolumab and pembrolizumab) have since been added to the guidelines. Based on the version 4.2021 guidelines, preferred regimens for mGC include fluoropyrimidines plus oxaliplatin or cisplatin (combined with trastuzumab for patients whose tumors overexpress HER2) or fluoropyrimidines plus oxaliplatin and nivolumab (for patients whose tumors express PD-L1 but not HER2) in the first-line setting and ramucirumab plus paclitaxel, fam-trastuzumab deruxtecan-nxki for patients whose tumors overexpress HER2, single-agent paclitaxel, single-agent docetaxel, fluoropyrimidine plus irinotecan, or single-agent irinotecan in the second-line setting.²

Despite the relatively limited number of chemotherapy regimens characterized as Category 1 in the first-line setting, many regimens are deemed appropriate, and optimal treatment strategies have yet to be defined. It stands to reason that a disease like mGC which carries a poor prognosis and has many reasonable treatment options is associated with very high degree of treatment heterogeneity.^3,4 This objective of study was to evaluate treatment patterns over time and to identify factors associated with treatment decisions to further understand the factors associated with treatment decision making for patients diagnosed with mGC in the US.

A retrospective observational study was conducted using deidentified records extracted from the IntelliDose chemotherapy order entry (COE) platform (IntrinsiQ Specialty Solutions, Frisco, TX). The data source is a nationwide commerciallyavailable COE system of cancer patients who receive systemic therapy at academic, private-practice and nonacademic hospital-based centers across the US. Eligible patients in this database were those who initiated treatment for mGC between 01/01/2006 and 11/30/2016, were aged ≥18 years and who received treatment at sites registered with IntelliDose for at least 2 years (to ensure continuity of longitudinal data).

Descriptive statistics included patient characteristics and frequencies of usage of specific chemotherapeutics with respect to treatment line over time. The distribution of baseline patient and physician characteristics across stage and specific treatment regimens was evaluated using Mantel-Haenszel or Pearson chi-square tests for categorical variables and Kruskal-Wallis or t-test or ANOVA for continuous variables. Treatment patterns for chemotherapy and for biologic/targeted therapies were evaluated from 2007 through 2016, including only those years with complete data in the database. Multivariable logistic regression was conducted to evaluate the association between patient and provider characteristics and use of selected chemotherapy regimens. Covariates included age, sex, type of practice setting, year of treatment initiation, geographic region of the practice, and practice size. SAS 9.4 (SAS Institute, Cary, NC) was used for all statistical analyses and significance was defined as P<.05.

Missing data are to be expected in real-world data sets. No imputation was made to account for incomplete data. However, patients with partial data were included in analyses for variables for which there were complete data (ie cases were not be fully excluded due to missing data).

A total of 4,276 mGC patients met eligibility criteria and were included in this study (Table 1). The median age was 64 years and 75.4% (n=3223) of the study cohort were male. Of eligible patients, 22.2% (n=950) were treated in an academic practice setting. The study cohort primarily received treatment in the Central (40.5%, n=1732) and East (41.5%, n=1772) regions of the US.

In the first-line setting, 23.0% received fluoropyrimidine + oxaliplatin (FOLFOX/CAPEOX), 14.8% received docetaxel + cisplatin or carboplatin + fluoropyrimidine (DCF), 14.0% received epirubicin + platinum + fluoropyrimidine [ECF/EOX], 12.7% received cisplatin or carboplatin + paclitaxel, 9.8% received cisplatin or carboplatin + fluoropyrimidine (CF), 7.5% received single-agent fluoropyrimidine, and 4.6% received a taxane (docetaxel or paclitaxel) as the chemotherapy backbone (with or without biologic or targeted agents) across the entire study period (Table 2). FOLFOX/CAPEOX use was lowest in 2007 at 4.2% and increased to 38.1% in 2016 while DCF use dropped from its peak of 28.2% in 2007 to 6.4% by 2016. ECF/EOX rose from 9.9% in 2007 to a peak of 20.5% in 2011, then dropped to 10.1% by 2016 (Figure 1, selected data labels added).Biologics were used among 10.7% of all patients initiating first-line therapy (Table 3).There were 213 unique treatment regimens observed among patients with mGC treated in the first-line setting. Trastuzumab was the most common biologic agent, received by 7.6% of patients in the first-line setting. First-line trastuzumab use increased from 1.0% in 2009 to 15.3% in 2016 (Figure 2, selected data labels added).

In the second-line setting, 14.7% received a taxane,14.7% received FOLFOX/CAPEOX, 12.6% received fluoropyrimidines, 8.1% received cisplatin or carboplatin + paclitaxel, and 10.9% received FOLFIRI as the chemotherapy backbone (with or without biologic or targeted agents) across the study period (Table 2). Biologic or targeted agents were used in 23.1% of all second-line starts. Trastuzumab and ramucirumab were the most commonly used biologics during the study period (12.2% and 6.5%, respectively). While chemotherapy patterns did not change in any clear direction in the second-line setting (Figure 3),the trends of second-line biologic/targeted agent use show use of trastuzumab beginning in 2009 and use of ramucirumab beginning in 2014 (Figure 4). There were 195 unique treatment regimens observed among patients with mGC treated in the second-line setting. Second-line trastuzumab use increased from 0% in 2008 to 22.8% in 2016 and second-line ramucirumab use increased from 0% in 2013 to 22.8% in 2016 (Figure 4).

Multivariable analyses show that older patients were less likely to receive DCF in the first-line setting. FOLFOX/ CAPEOX was also more likely to be used in the first-line setting in recent years; whereas DCF was less likely to be used. There were no significant differences in the use of cisplatin or carboplatin + paclitaxel by year of treatment initiation. The full results of significant factors in the multivariable analyses are presented in Table 4.

Treatment heterogeneity remains highly prevalent in the mGC chemotherapy landscape. This study observed a very high number of different regimens utilized across lines of therapy, and leads to the question of potential value of treatment pathways in this disease. Investigators at the US Oncology Network recently examined the impact of implementing pathways in gastric cancer and observed increased concordance with NCCN guidelines and reduced heterogeneity.5 Possible reasons for the high level of treatment heterogeneity observed in mGC were explored in this study. Clinical factors, such as patient age, were identified as significant predictors of certain treatment regimens. Specifically, older age was associated with lower utilization of DCF, likely due to the more favorable toxicity profile of doublet versus triplet chemotherapy regimens. Treatment guidelines have evolved accordingly; ECF/EOX and DCF had both previously been included as Category 1 preferred regimens in NCCN guidelines, but in version 1.2017, were reassigned a Category 2A and 2B designation, respectively. FOLFOX/CAPEOX, on the other hand, has consistently remained a Category 2A preferred regimen.² These changes in practice guidelines are reflected in the observed patterns of care, which also demonstrated an increasing use of FOLFOX with decreasing use of ECF/EOX and DCF.

The use of biologic therapy increased to more than 20% in the first- and second-line settings, respectively. Trastuzumab for patients with HER2+ mGC was approved for use by the U.S. Food and Drug Administration (FDA) in 2010, and ramucirumab (alone or in combination with paclitaxel after disease progression) was approved in 2014. Initiation and rapid increase in the use of both drugs were observed following these FDA approvals. Both agents have demonstrated statistically significant improvements in overall survival and are included in NCCN guidelines as Category 1 preferred regimens.^2,6-8 Trastuzumab is only appropriate for the subset of patients who overexpress HER2 (12-23% of patients) and is recommended in the first-line setting.² During the period of this study, trastuzumab use was observed among 15% of patients treated in the first line and more than 22% in the second-line setting. Since the time period of this study, the continued use of trastuzumab beyond progression has been shown to be of no benefit to patients.⁹ Ramucirumab use was first observed in 2014 and was used by more than 22% of patients in both 2015 and 2016 in the second-line setting.

There are several potential limitations that must be taken into consideration when reviewing these findings. Data on race, insurance coverage, and socioeconomic status were unavailable for inclusion. Moreover, the database has no information about patient survival or adverse events. Therefore, influence of the toxicity profiles of specific regimens could not be directly explored in this study due to the lack of adverse event data within this real-world dataset. Nonetheless, we were able to find important trends in treatment patterns in the US. The patient cohort may not be representative of the larger mGC patient population treated in the US, as these data are recorded from a single source. However, the 2012 American Society of Clinical Oncology (ASCO) National Census of Oncology Practices found the distribution of practice settings relatively consistent with outpatient cohort (ASCO practice survey: academic=9.5%, private practice=56%, community hospital=24% vs. our study: academic=8%, private practice=54% and community hospitals=38%), suggesting that the patient cohort in this study may indeed be representative of the larger US mGC patient population. Lastly, the time period of study observed little use of immune checkpoint inhibitors. Future studies may be warranted to further investigate the use of these agents following their inclusion in national guidelines. 

The potential for data misclassification always represents a concern. This study relied on data from a COE system designed for safety purposes to ensure accurate entry of drugs, doses, patient characteristics, disease indication, and stage in a manner readily available for quality audits and analysis. To further ensure data quality, nurses employed by IntrinsiQ routinely reviewed data from all sites to assess for data entry inaccuracies or deviation from standard practice. Finally, the study period preceded the approval of immunotherapy for gastric cancer by the FDA, so we were not able to assess the use of those agents in gastric cancer.

The changes in treatment choice over time and associated with patient clinical factors suggest the central role of NCCN guidelines in shaping the standard of care for these patients, and initiation and rapid increase in the use of both trastuzumab and ramucirumab were observed following their FDA approvals in 2010 and 2014, respectively. There remains significant heterogeneity in mGC treatment and there is a need to emphasize the use of evidence-based care for these patients.

Authors: Thomas A. Abrams, MD¹; Gary Meyer, PhD²; Lisa M. Hess,PhD³;, Sui Zhang, MS, MPP¹, Astra M. Liepa, PharmD³; Yajun Zhu, MS³; Jeffrey A. Meyerhardt, MD, MPH¹ ; Deborah Schrag, MD, MPH¹; Peter C. Enzinger, MD¹; Charles S. Fuchs⁴

Affiliations: ¹Dana-Farber Cancer Institute, Boston, MA
²IntrinsiQ Specialty Solutions, a part of AmerisourceBergen, Frisco, TX
³ Eli Lilly and Company, Indianapolis, IN
⁴Yale Cancer Center & Smilow Cancer Hospital, New Haven, CT

Correspondence: Thomas A. Abrams, MD
Phone:+1 (617) 632-6932
Email: thomas_abrams@dfci.harvard.edu

Disclosures: This work was supported by a grant from Eli Lilly and Company and with full consent and participation of IntrinsiQ Specialty Solutions, a part of AmerisourceBergen (Frisco, TX), which received funding for the provision of data for this study. The contents of this article are solely the responsibility of the authors and do not necessarily represent views of Eli Lilly and Company or IntrinsiQ Specialty Solutions.

Acknowledgements: The research presented in this manuscript is original. A poster reflecting these data was presented at the ASCO 2018 Gastrointestinal Cancers Symposium in San Francisco, CA on January 19, 2018.