Summary of Study Characteristics
The detailed characteristics of the included SR21 (), primary clinical studies (RCTs)22–35 (), economic studies36–40 (), and guidelines41–47 () are provided in Appendix 2.
Study Design
The SR21 included 26 studies comparing ondansetron with other 5-HT3 RA treatments, of which 6 RCTs compared ondansetron with palonosetron. The RCTs included in the SR were published between 2003 to 2013.
The additional 12 included RCTs published in 14 reports comprised 7 open-label22,23,25–27,31,34 and 5 double-blind24,28–30,32,33,35 trials. Ten RCTs were parallel22–25,27–30,32–35 and 2 were cross-over26,31 in design. Sample size calculation was performed and reported in 9 RCTs.23–25,27,29,31,32,34,35 The efficacy results were analyzed using intention-to-treat approach in 2 RCTs28,31 and per-protocol analysis or not was reported in the other 10 studies.22–27,29,30,32–35
Of the 5 included economic studies, 4 were cost-utility analysis36,37,39,40 and 1 was a cost-effectiveness analysis.38
The cost-utility analysis by Botteman et al. (2020)36 was conducted using the efficacy data from a phase III noninferior RCT by Zhang et al. (2018)29 to determine the cost-effectiveness of NEPA relative to the granisetron-aprepitant regimen for HEC. The analyses were performed from the US health care perspective, with a time horizon of 5 days. The utilities values of 0.90, 0.70, and 0.24 were assigned for the outcomes of complete protection (CP), complete response (CR), and incomplete response (IR), respectively. The costs of antiemetic prophylaxis, rescue medications, and medical costs of CINV-related events were assigned into the analysis based on usage observed in the trial. Costs were adjusted at 2018 US dollars.
The cost-utility analysis by Kashiwa and Matsushita (2019)37 was conducted using the efficacy data of a phase III RCT (TRIPLE study) by Suzuki et al. (2016)35 to assess the cost-effectiveness of a triple regimen of palonosetron relative to granisetron for cisplatin-containing HEC. The analyses were performed from the Japanese health care payer perspective, with a time horizon of 5 days. The utilities values of 0.90, 0.70, and 0.20 were assigned for the outcomes of CP, CR, and IR, respectively. Costs considered in the analyses included direct medical costs associated with CINV prevention and medical fees incurred by CINV. Costs were adjusted at 2018 US dollars (US$1 = 112.17 Japanese yen [JPY]).
The cost-effectiveness analysis by Shimizu et al. (2018)38 was also conducted using the efficacy data of a phase III RCT (TRIPLE study) by Suzuki et al. (2016)35 to assess the cost-effectiveness of a triple regimen of palonosetron relative to granisetron for cisplatin-containing HEC. The analyses were performed from the Japanese health care payer perspective, with a time horizon of 5 days. The costs of drugs and total medical costs were considered in the analysis. Costs were adjusted at 2018 US dollars (US$1 = 110.57 JPY).
The cost-utility analysis by Du et al. (2017)39 was conducted using the efficacy data of 2 pivotal phase III RCTs to compare the cost-effectiveness among 3 5-HT3 RAs (i.e., palonosetron, ondansetron, and granisetron) in the presence of DEX for HEC. The analyses were performed from the Chinese health care perspective, with a time horizon of 5 days. The utilities values of 9.02, 7.74, and 2.28 were assigned for the outcomes of CR, nausea but not receive medication, and failure, respectively. Only direct medical costs (i.e., antiemetic drugs and rescue drugs) were considered in the analyses. Costs were adjusted at 2014 US dollars.
The cost-utility analysis by Restelli et al. (2017)40 was conducted using the efficacy data of 3 RCTs48–50 to determine the cost-effectiveness of NEPA-DEX relative to other comparators for HEC or MEC. The analyses were performed from the Italian National Health Service perspective, with a time horizon of 5 days. The utilities values of 0.77, 0.60, and 0.26 were assigned for the outcomes of CP, CR, and IR, respectively. Direct medical costs, costs management of adverse events (AEs), and costs for the management of CINV episodes were considered in the analysis. Costs were adjusted at year 2016, in Euros.
All 5 included guidelines were updated versions of the previous American Society of Clinical Oncology (ASCO) guideline,41 National Comprehensive Cancer Network (NCCN) guideline,42 Cancer Care Ontario (CCO) guideline,43 Pediatric Oncology Group of Ontario (POGO) guideline,44 and Multinational Association of Supportive Care in Cancer/ European Society of Medical Oncology (MASCC/ESMO) guideline.45–47 All guidelines were developed to provide recommendations on the use of antiemetics for the prevention of CINV. A systematic literature review search was conducted for all included guidelines. The quality of evidence and the strength of recommendations were assessed and reported in 4 guidelines: ASCO,41 NCCN,42 POGO,44 and MASCC/ESMO.45–47 In the CCO guideline,43 the quality of evidence was assessed, but the level of evidence and the strength of recommendations were not provided for each recommendation. Recommendations in all guidelines were developed by expert panels and the guidelines were reviewed by external reviewers and published either on their websites or in peer-reviewed journals.
Country of Origin
The SR21 was conducted by authors from Brazil and was published in 2016.
The RCTs were conducted by authors in Egypt,22 the Netherlands,23 Japan,24,26,31,35 India,25,27 China,28–30 Korea,34 the US and multiple countries in Latin America, Western and Eastern Europe, and Russia.32,33 Two RCTs were published in 2021,22,23 1 in 2020,24 2 in 2019,25,26 3 in 2018,27–29 and 4 in 2016.31,32,34,35 Four RCTs were designed to test the noninferiority of palonosetron compared to ondansetron23,27,32 or to granisetron.29 The noninferiority margin was set at −10%,29 −15%27,32 or −20%.23 Noninferiority was demonstrated if the lower limit of the confidence interval (CI) for the difference between palonosetron and its comparator (ondansetron or granisetron) in the incidence of primary outcome was greater than the noninferiority margin.
The economic studies were conducted by authors from US,36 Japan,37,38 China,39 and Italy.40 The studies were published in 2020,36 2019,37 2018,38 and 2017.39,40
The guidelines were conducted by authors from US,41,42 Canada,43,44 and multiple countries including Canada, the US, and European countries.45–47
Patient Population
Patients in the RCTs included in the SR21 were adults with various cancer types (e.g., breast, lung, bladder, colon, rectum, gastric, lymphoma, leukemia, other) who were scheduled to receive HEC (3 RCTs) or MEC (3 RCTs). The mean age ranged between 52 years and 56 years, and the percent of females ranged between 36% and 100% (breast cancer).
Eight RCTs22–24,26,29–31,34,35 involved adult cancer patients with mean ages ranging from 49 to 68 years, and the proportion of females varied from 20% to 100% (breast cancer). Five RCTs included patients naive to chemotherapy who were scheduled to receive HEC,22,24,29–31,35 1 RCT included patients naive to chemotherapy who were scheduled to receive MEC,23 1 RCT included patients naive to chemotherapy who were scheduled to receive HEC or MEC,26 and 1 RCT included patients with or without previous chemotherapy who were scheduled to receive MEC.34 The types of cancer were mainly lung,22,26,29,30,35 breast,22,24 colon and rectum,23 and stomach.31 One RCT did not report cancer type.34
There were 4 RCTs25,27,28,32,33 involving pediatric patients with mean age ranging from 5 to 8 years and the proportion of females ranged between 28% to 52%. Two RCTs included patients naive to chemotherapy who were scheduled to receive HEC or MEC,25,27 1 RCT included patients with or without previous chemotherapy who were scheduled to receive HEC or MEC,32,33 and 1 RCT included patients with or without previous chemotherapy who were scheduled to receive HEC.28 The types of cancer included both hematological cancer and solid tumour.
Patients in all economic studies were adults with cancer, naive to chemotherapy, who were scheduled to receive HEC;36–39 or adult cancer patients naive to chemotherapy who were scheduled to receive HEC or MEC.40
All included guidelines41–47 were developed for health care providers involved in the treatment and care for cancer patients including oncologists, pharmacists, and nurses. The target population of the ASCO guideline41 and the MASCC/ESMO guidelines45–47 are adult patients receiving HEC or MEC, and pediatric patients receiving HEC or MEC. The target population of the NCCN guideline42 are adult patients receiving HEC or MEC parenteral anticancer drugs and adult patients receiving HEC or MEC oral anticancer drugs. The target population of the CCO guideline43 are adult patients receiving HEC or MEC, a single day of IV chemotherapy or a multiple day of IV chemotherapy. The target population of the POGO guideline44 is pediatric patients receiving HEC or MEC.
Interventions and Comparators
The SR21 included 6 RCTs that compared ondansetron with palonosetron. Ondansetron doses varied from 32 mg administered orally, or from 8 mg to 32 mg administered intravenously. The palonosetron dose was 0.25 mg administered either orally or IV. Meta-analysis was performed and subgroup analysis was conducted.
Among RCTs involving adult patients, 2 RCTs compared a dual regimen of palonosetron-DEX versus granisetron-DEX,22,34 1 RCT compared palonosetron-DEX versus ondansetron-DEX versus ondansetron-DEX-metoclopramide,23 3 RCTs compared a regimen of palonosetron-aprepitant or fosaprepitant-DEX versus granisetron-aprepitant or fosaprepitant-DEX,24,26,35 1 RCT compared NEPA oral (netupitant 300 mg and palonosetron 0.5 mg)-DEX versus granisetron-aprepitant-DEX,29 and 1 RCT compared palonosetron-DEX versus granisetron-aprepitant-DEX,31
Among RCTs involving pediatric patients, 3 RCTs compared dual regimens of palonosetron-DEX versus ondansetron-DEX,25,28,32 and 1 RCT compared a dual regimen of palonosetron-DEX versus ondansetron-DEX for MEC and a triple regimen of palonosetron-fosaprepitant-DEX versus of ondansetron-fosaprepitant-DEX for HEC.27
The cost-utility analysis by Botteman et al. (2020)36 assessed the cost-effectiveness of NEPA-DEX relative to granisetron-aprepitant-DEX in patients following HEC.
The cost-utility analysis by Kashiwa and Matsushita (2019)37 and the cost-effectiveness analysis by Shimizu et al. (2018)38 determined the cost-effectiveness of a triple regimen of palonosetron-aprepitant-DEX versus granisetron-aprepitant-DEX in patients following HEC.
The cost-utility analysis by Du et al. (2017)39 estimated which of the 3 treatment strategies consisting of palonosetron-DEX, ondansetron-DEX, and granisetron-DEX was the most cost-effective option in patients following HEC.
The cost-utility analysis by Restelli et al. (2017)40 estimated the cost-effectiveness of NEPA-DEX compared with other regimens such as palonosetron-aprepitant-DEX, palonosetron-fosaprepitant-DEX, ondansetron-aprepitant/DEX, and ondansetron-fosaprepitant-DEX in patients following HEC or MEC.
All included guidelines41–47 considered the efficacy and safety of antiemetic drugs including 5-HT3 RA and NK-1 RA used concomitantly with DEX, which were formulated in different regimens for HEC or MEC.
Outcomes
The efficacy outcomes examined in the SR21 were acute nausea, acute vomiting, delayed nausea, and delayed vomiting. The AEs that were assessed included headache, constipation, diarrhea, and dizziness. The acute phase was defined as 0 to 24 hours after chemotherapy and the delayed phase was from more than 24 hours to 120 hours. Follow-up periods were not reported.
The clinical outcomes considered in the RCTs included CR, total control (TC), complete control (CC), CP, nausea, vomiting, and no use of rescue antiemetic medication. CR was defined as no vomiting and no use of antiemetic medication. CP was defined as no nausea and no vomiting. TC was defined as no vomiting, no use of antiemetic medication and no nausea. CC was defined as no vomiting, no use of antiemetic medication and no more than mild nausea. These outcomes were assessed for acute (within 24 hours), delayed (24 to 120 hours), and overall (0 to 120 hours) period after completion of chemotherapy. Other outcomes considered in the RCTs were the MASCC Antiemetic Tool (MAT) questionnaire, quality of life (QoL) assessed using the Functional Living Index-Emesis (FLIE), and AEs. The MAT questionnaire was a validated tool and an 8-item scale for the assessment of acute and delayed CINV that is completed once per cycle of chemotherapy. The FLIE questionnaire consists of 9 nausea-specific items and 9 vomiting-specific items. Responses were marked on 100 mm visual analogue scale (VAS), with anchors of 1 and 7. A total FLIE score of higher than 108 was considered as “no or minimal impact on daily life.” All RCTs, except one (follow-up of 3 days),34 had a follow-up period of 5 days (120 hours).
The primary outcomes in the cost-utility analysis by Botteman et al. (2020)36 were net monetary benefit (NMB) and the probability that NEPA/DEX is cost-effective versus granisetron-aprepitant-DEX. NMB was calculated with the formula: NMB = quality-adjusted life-day (QALD) difference / 365.25 × $25,000 – cost difference. The willingness-to-pay (WTP) per quality-adjusted life-year (QALY) gained threshold was set at $25,000. A positive NMB suggests that NEPA is cost-effective at the $25,000 per QALY threshold. The higher the NMB, the more cost-effective NEPA is. A sensitivity analysis was conducted using a 1-way and probabilistic sensitivity analysis approach to confirm the robustness of the base-case results.
In the cost-utility analysis by Kashiwa and Matsushita (2019),37 cost-effectiveness was calculated from the costs incurred in antiemetic therapy and QALYs for 5 days. The incremental cost-effectiveness ratio (ICER) of the base case was calculated by dividing the incremental cost between regimens divided by incremental QALYs between regimens. The WTP threshold was 5,000,000 JPY per QALY (US$44,575 per QALY). One-way and probabilistic sensitivity analyses were conducted to assess the uncertainty and robustness of the model.
In the cost-effectiveness analysis by Shimizu et al. (2018),38 the cost-effectiveness ratio was calculated by dividing the mean cost of antiemetic used in each group by the number of CR. The ICER was calculated as the difference in mean cost between groups divided by the difference in CR rates between groups. One-way sensitivity analysis of branded and generic drugs as rescue medication was carried out to calculate the ICER range.
In the cost-utility analysis by Du et al. (2017),39 the ICER was estimated for palonosetron or ondansetron compared with granisetron. The WTP threshold was set at US$22,515. One-way and probabilistic sensitivity analyses were conducted to reflect the uncertainty and robustness of the model.
In the cost-utility analysis by Restelli et al. (2017),40 the ICER of NEPA-DEX was compared with palonosetron-aprepitant-DEX, palonosetron-fosaprepitant-DEX, ondansetron-aprepitant-DEX, and ondansetron-fosaprepitant-DEX. A 1-way sensitivity analysis was conducted to test the robustness of the results.
All included guidelines41–47 considered evidence-based on efficacy and safety outcomes of antiemetic drugs for the prevention of CINV, for the development of the recommendations.
Summary of Critical Appraisal
The detailed quality assessments of the included SR21 (), RCTs22–35 ( and ), economic studies36–40 (), and guidelines41–47 () are presented in Appendix 3.
The SR21 was explicit in its objective and inclusion criteria for the review and selection of study design for inclusion, and included a comprehensive literature search strategy. Study selection was performed in duplicate, but it was unclear if data extraction was performed in duplicate. The SR did not report whether a protocol had been published before the conducting of the review. The SR also did not report the sources of funding of the studies included in their review, nor did they provide a list of excluded studies. The characteristics of the included studies were described in adequate detail. A modified Jadad scale and the Cochrane risk of bias (RoB) tool were used to assess the quality and RoB of the included studies. Meta-analysis was performed to combine the results and a subgroup analysis was conducted to assess the potential impact of RoB on the results. Statistical heterogeneity was observed and discussed. Publication bias was not investigated due to the small number of studies. Conflicts of interest were declared. Overall, the SR was of acceptable methodological quality.
All included RCTs22–35 were explicit in reporting (i.e., clearly described the objective of the study, the main outcomes, the characteristics of the participants, the interventions, differences in baseline characteristics between groups, and the main findings of the study). All RCTs22–35 provided estimates of the random variability (e.g., standard deviation or 95% CI) in the data of the main outcomes and actual P values for main outcomes. Of the included RCTs, 222,26 did not report AEs related to treatment drugs. As 4 RCTs22,26,27,31 were conducted with relatively small sample sizes (range from 70 to 116), it was not applicable to determine if the participants were representative of the entire population from which they were recruited. However, the treatment settings in all included RCTs were representative of the treatment received by most of the patients. Seven RCTs22,23,25–27,31,34 were open-label and 5 RCTs24,28–30,32,33,35 were double-blind, of which 229,30,32,33 were double-blind/double-dummy. The intervention and comparator groups in all included RCTs had the same follow-up. Appropriate statistical tests were used to assess the main outcomes, which were accurately measured. Patients in all intervention groups were recruited from the same population and over the same time period. Allocation concealment was only reported in 1 RCT.32,33 Analysis for efficacy and safety outcomes was performed using intention-to-treat (ITT) analysis in 2 RCTs.28,31 Sample size was determined in 9 RCTs23–25,27,29–35 and not reported in the other 3.22,26,28 Overall, the methodological quality of the included RCTs was moderate to high.
All included economic studies36–40 clearly stated the objectives, the economic importance of the research questions, the rationale for choosing the alternative comparators, the viewpoint of the analysis, and the type of economic evaluation that was conducted. Three studies37,39,40 justified the choice of form of economic evaluation in relation to the questions addressed. For data collection, all economic studies clearly stated the sources of effectiveness estimates, with details of the design and findings of those studies, the primary outcome measures for the economic evaluation, the methods to value benefits, the methods for the estimation of quantities and unit costs, currency and price data, and details of the model used (except for 2 studies,36,38 which did not have a model). For the analysis and interpretation of results, all economic studies clearly stated the time horizon of costs and benefits, details of statistical tests and CIs, and the approach to sensitivity analysis. All studies provided justification for the choice of variables for sensitivity analysis and the ranges over which the variables were varied. All studies reported incremental analysis and presented major outcomes in a disaggregated, as well as aggregated, form. The conclusions in all the studies were based on the data reported and were accompanied by the appropriate caveats. Overall, the included economic studies were of moderate to high methodological quality in study design, data collection, and analysis and interpretation of results.
All included guidelines41–47 were explicit in their scope and purpose (i.e., objectives, health questions, and populations), and had clear presentation (i.e., specific, and unambiguous recommendations, different options for management of the condition or health issue, and easy to find key recommendations). Regarding stakeholder involvement, all guidelines clearly defined target users and the development groups; however, it was unclear if the views and preferences of the patients were sought. For rigour of development, all guidelines reported details of systematic searches for evidence, criteria for selecting evidence, explicit link between recommendations and the supporting evidence, and methods of formulating the recommendations. All guidelines considered health benefits, side effects, and risks in formulating the recommendations; were peer-reviewed before publication; and provided a procedure for updating. All guidelines, except the CCO guideline,43 assessed and reported the strength of its recommendations. For applicability, the guidelines were explicit in facilitators and barriers to application, advice and/or tools on how the recommendations can be put into practice, resource implications, and monitoring and or auditing criteria. For editorial independence, all guidelines reported that the funding bodies had no influence on the content of the guidelines. The competing interests of the guideline development group members were reported. Overall, all included guidelines were of high methodological quality.
Limitations
Although there has been a large body of clinical evidence within the past 5 years regarding the clinical effectiveness of palonosetron versus ondansetron and granisetron in the prevention of CIVN, the included SR and RCTs had several limitations. The SR21 was published in 2016 and its included studies for the comparison of palonosetron versus ondansetron were published between 2003 and 2013; hence, the evidence was quite outdated relative to the included RCTs. Studies included in the SR were heterogeneous in their treatment regimens of the interventions, sample size, type of cancer, type of chemotherapy, and concomitant used of corticosteroids. One of the limitations of the included RCTs was that treatment regimens including the dosage of the intervention drugs (i.e., palonosetron, ondansetron, and granisetron), schedule and mode of administration, and the use of concomitant medications such as NK1 RA and DEX, varied among studies even within the same type of chemotherapy and therefore would generate different results. Another limitation was that the primary outcome varied among studies and it was unclear how it was selected. It was also unclear how the noninferiority margin was determined in the noninferiority studies. Seven RCTs22,23,25–27,31,34 had open-label designs, which may allow the analysis to be vulnerable to detection bias. As 4 RCTs22,26,27,31 were conducted with relatively small sample sizes, the non-significant differences in certain outcomes between groups may be due to the lack of power. Patients’ comorbidities, concomitant addition medications or current home medications, could interfere with the medication used and were not assessed in the included studies. Also, subgroup analysis by patient risk factors of emesis such as age, sex, history of morning sickness, anxiety, and expectations of nausea and vomiting was not performed.
One of the limitations in the included economic evaluations36–40 was that the costs and benefits for treatment were limited to short time periods (i.e., 120 hours). All studies just focused on the effect of antiemetics within 120 hours in the first cycle of chemotherapy; therefore, the economic evaluations could not track the additional use of chemotherapy and additional use of antiemetics to prevent CINV. In the studies conducting cost-utility analysis, the utility values were based on data measured in other countries, which have different health care systems. The cost-utility analysis was based on a clinical efficacy trial so that the results may have high internal validity; however, the degree of external validity may be limited when extrapolating the results to different populations. Incidence and duration of treatment-related AEs were obtained from well-controlled trials that may not reflect real-word data. The cost-utility analysis did not include utility values associated with AEs of antiemetics; therefore, the costs associated with antiemetic prophylaxis may be underestimated. The included economic studies were conducted in countries other than Canada; therefore, the results have limited generalizability to the Canadian context (i.e., the Canadian health care system).
There were no significant methodological limitations of all included guidelines, except that the strength of recommendations in the CCO guideline43 was not graded.