1019672 research-article20212021 TAM0010.1177/17588359211019672Therapeutic Advances in Medical OncologyP Jimenez-Fonseca, A Carmona-Bayonas Therapeutic Advances in Medical Oncology Original Research External validity of clinical trials with Ther Adv Med Oncol 2021, Vol. 13: 1­–13 diverse trastuzumab-based chemotherapy DOI: 10.1177/ https://doi.org/10.1177/17588359211019672 https://doi.org/10.1177/17588359211019672 17588359211019672 regimens in advanced gastroesophageal © The Author(s), 2021. Article reuse guidelines: sagepub.com/journals- adenocarcinoma: data from the permissions AGAMENON-SEOM registry Correspondence to: Alberto Carmona- Paula Jimenez-Fonseca, Alberto Carmona-Bayonas, Alba Martinez-Torron, Maria Alsina, Bayonas Ana Custodio, Olbia Serra, Diego Cacho Lavin, María Luisa Limón, Tamara Sauri, Medical Oncology Department, Hospital Flora López, Laura Visa, Mónica Granja, Nieves Martínez Lago , Virginia Arrazubi, Universitario Morales Rosario Vidal Tocino, Raquel Hernandez, Gema Aguado, Juana María Cano, Meseguer, Calle Marqués de los Vélez, s/n, Murcia, Alfonso Martín Carnicero, Monserrat Mangas, Paola Pimentel, Ana Fernández Montes, 30007, Spain alberto. Ismael Macias Declara, Federico Longo, Avinash Ramchandani, Marta Martín Richard, carmonabayonas@gmail. Alicia Hurtado, Aitor Azkarate, Carolina Hernández Pérez, Raquel Serrano and Javier com Gallego on behalf of the AGAMENON-SEOM study group Paula Jimenez-Fonseca Medical Oncology Department, Hospital Universitario Central de Abstract Asturias, ISPA, Oviedo, Background: Trastuzumab combined with cisplatin and fluoropyrimidines, either capecitabine Spain Alba Martinez-Torron or 5-fluorouracile (XP/FP), is the standard first-line treatment for advanced, HER2-positive, Pharmacy Department, gastric cancer patients based on the ToGA trial. Despite the lack of phase III trials, many Hospital Universitario Marqués de Valdecilla, clinicians administer trastuzumab with alternative regimens. One meta-analysis suggests that Santander, Spain substituting cisplatin for oxaliplatin might lead to greater efficacy and less toxicity. Maria Alsina Medical Oncology Methods: 594 patients with HER2-positive gastroesophageal adenocarcinoma were recruited Department, Complejo from the AGAMENON-SEOM registry. The objective was to evaluate the external validity of Hospitalario de Navarra, Pamplona Vall Hebron clinical trials with chemotherapy and trastuzumab. Institute of Oncology (VHIO), Barcelona, Spain Results: The regimens used in at least 5% of the patients were XP (27%), oxaliplatin and Ana Custodio capecitabine (CAPOX) (26%), oxaliplatin and 5-fluorouracil (FOLFOX) (14%), FP (14%), triplet Medical Oncology with anthracycline/docetaxel (7%), and carboplatin-FU (5%). Median exposure to trastuzumab Department, Hospital Universitario La was longer with FOLFOX (11.4 months, 95% CI, 9.1–21.0) versus ToGA regimens (7.5, 6.4–8.5), Paz, CIBERONC CB16/12/00398, Madrid, p < 0.001. Patients with HER2-IHC 3+ cancers had higher response rates than those with IHC Spain 2+/FISH+, odds-ratio 1.97 (95% CI, 1.25–3.09). The results achieved with CAPOX–trastuzumab Olbia Serra Medical Oncology were comparable to those attained with ToGA regimens. FOLFOX–trastuzumab was superior Department, Catalan to ToGA schemes in terms of overall survival (OS), with a greater magnitude of effect in IHC Institute of Oncology, L’Hospitalet, Spain 2+/FISH+ tumors (HR 0.47, 0.24–0.92) compared with IHC 3+ (HR 0.69, 0.49–0.96), and in Diego Cacho Lavin diffuse (HR 0.37, 0.20–0.69) versus intestinal-type tumors (HR 0.76, 0.54–1.06). Medical Oncology Department, Hospital Conclusion: We have updated the external validity of clinical trials with trastuzumab in Universitario Marqués first-line treatment of gastric cancer. Our data confirm the comparable outcomes of ToGA de Valdecilla, IDIVAL, Santander, Spain regimens and CAPOX–trastuzumab in clinical practice and point toward a possible benefit of María Luisa Limón FOLFOX–trastuzumab, contingent on the subtypes typically less sensitive to trastuzumab, to Medical Oncology Department, Hospital be confirmed in clinical trials. Universitario Virgen del Rocío, Sevilla, Spain Tamara Sauri Keywords: chemotherapy, external validity, gastric cancer, HER2, oxaliplatin, trastuzumab Medical Oncology Department, Hospital Clinic, IDIBAPS, Received: 13 February 2021; revised manuscript accepted: 29 April 2021. Barcelona, Spain journals.sagepub.com/home/tam 1 Creative Commons Non Commercial CC BY-NC: This article is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 License (https://creativecommons.org/licenses/by-nc/4.0/) which permits non-commercial use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage). Therapeutic Advances in Medical Oncology 13 Flora López Introduction profile, although it was unclear whether this Medical Oncology Department, Hospital Advanced gastric cancer often exhibits chromo- depended on patient- or cancer-related factors or Universitario Doce de somal anomalies that can be acted upon through on treatment duration.10 Trastuzumab associated Octubre, Madrid, Spain receptor tyrosine kinase (RTK)-directed thera- with anthracycline or docetaxel-based triplet Laura Visa Medical Oncology pies.1 Overexpression of human epidermal growth increased toxicity without any gain in efficacy. Department, Hospital factor receptor 2 (HER2) comprises one of the Nevertheless, the data in favor of regimens with Universitario El Mar, Barcelona, Spain most recurring alterations (9–36%).2,3 In 2010, oxaliplatin derive from three small phase II trials Mónica Granja the phase III randomized controlled trial (RCT) that, all told, amount to 140 individuals, and Medical Oncology Trastuzumab for Gastric Cancer (ToGA) dem- yield superior survival endpoints to those gener- Department, Hospital Universitario Clínico onstrated the therapeutic effect of adding trastu- ally seen in larger RCTs (median OS rates around San Carlos, Madrid, zumab to six cycles of cisplatin and 20 months).11–13 This leaves the door open to the Spain Nieves Martínez Lago fluoropyrimidine (capecitabine, fluorouracil) in existence of selection factors that could affect the Medical Oncology patients with cancer of the stomach or of the gas- phase II trial outcomes.14 Therefore, the indirect Department, troesophageal junction with HER2 amplification comparison between ToGA and the pooled esti- Complejo Hospitalario Universitario de A or overexpression (HER2+).4 Patients who mates of these small, heterogeneous studies (e.g. Coruña, A Coruña, Spain received trastuzumab reached longer overall sur- different eligibility criteria) implies that this kind Virginia Arrazubi Medical Oncology vival (OS), 13.8 versus 11.0 months, [hazard ratio of causal inference runs the risk of bias. Among Department, Complejo (HR) 0.74, 95% confidence interval (CI), 0.60– the limitations, this method impedes adjusting for Hospitalario de Navarra, IdiSNA (Navarra 0.91] and impairment of their quality of life covariates, controlling for selection or recruit- Institute for Health delayed.5 In the subgroup with immunohisto- ment bias; similarly, the effect on the overall Research), Pamplona, Spain chemistry (IHC) 2+ scoring and fluorescence in response rate (ORR) or progression-free survival Rosario Vidal Tocino situ hybridization (FISH)-positive tumors or IHC (PFS) is unknown.15 There are currently no Medical Oncology 3+ tumors, OS was 16.0 months. RCTs that compare ToGA versus trastuzumab Department, Complejo Asistencial Universitario with oxaliplatin-based regimens. de Salamanca-IBSAL, The European Medicines Agency approval of this Salamanca, Spain combination in 2010, notwithstanding many With this premise, this study of the AGAMENON- Raquel Hernandez Medical Oncology questions remaining unanswered, including opti- SEOM registry seeks to evaluate the external Department, Hospital mal chemotherapy duration or the most suitable validity of clinical trials with trastuzumab and Universitario de Canarias, Tenerife, backbone for it. Thus, several subsequent trials chemotherapy in HER2+ gastroesophageal ade- Spain have suggested that bi-weekly oxaliplatin-based nocarcinoma, analyze the results of the different Gema Aguado regimens are better tolerated.6,7 Instead of admin- regimens, and probe into associated predictive Medical Oncology Department, Hospital istering cisplatin- and fluoropyrimidine-based factors. Universitario Gregorio treatments (ToGA regimens), many clinicians Marañón, Madrid, Spain Juana María Cano tend to incorporate trastuzumab to the standard Medical Oncology first-line regimen in their department and tailored Materials and method Department, Hospital General Universitario to the patient’s specific characteristics (e.g. de Ciudad Real, Ciudad CAPOX, FOLFOX, or triplets). Similarly, there Patients and study design Real, Spain is no consensus in clinical guidelines with respect The study population is from the AGAMENON- Alfonso Martín Carnicero to chemotherapy regimen, duration, or the best SEOM gastroesophageal cancer registry, in which Medical Oncology maintenance strategy. While the National 34 Spanish centers currently participate.3,16–25 Department, Hospital San Pedro, Logroño, Comprehensive Cancer Network (NCCN) Eligibility criteria for this analysis include age Spain endorses the addition of trastuzumab to any >18 years; histologically confirmed, inoperable, Monserrat Mangas chemotherapy regimen, ESMO guidelines rec- locally advanced, recurrent, or metastatic adeno- Medical Oncology Department, Hospital ommend using cisplatin and fluoropyrimidine in carcinoma of the distal esophagus, stomach, or Galdakao-Usansolo, HER2+ cancers.8,9 gastroesophageal junction; HER2+ status, and Usansolo, Spain having received at least one cycle of polychemo- Paola Pimentel Medical Oncology A meta-analysis with 557 patients from 15 pro- therapy (two or more cytotoxic drugs) associated Department, Hospital spective or retrospective studies indirectly com- with trastuzumab. HER2 status was determined General Universitario Santa Lucía, Cartagena, pared OS of regimens alternative to ToGA.10 Ter locally at the participating centers by means of Spain Veer et al. concluded that oxaliplatin-based dual- IHC scoring and/or FISH as part of regular clini- Ana Fernández Montes Medical Oncology agent schemes (FOLFOX or CAPOX) associated cal practice. HER2+ overexpression was defined Department, Complejo with trastuzumab were more efficacious than as a +3 IHC score or +2 accompanied by FISH Hospitalario de Orense, Orense, Spain ToGA regimens (median OS 20.7 versus (FISH+). Those participants who had completed 16 months, HR 0.75, 95% CI, 0.59–0.99), which perioperative or adjuvant treatment in the pre- they attributed to its more favorable toxicity ceding 6 months and those HER2+ cancers not 2 journals.sagepub.com/home/tam P Jimenez-Fonseca, A Carmona-Bayonas et al. treated with trastuzumab in first-line treatment weekly) administered with respect to the planned Ismael Macias Declara were excluded. amount for each schedule. Multivariable models Medical Oncology included nine variables, selected on the basis of Department, Hospital The database is managed via an online platform theoretical criteria or experience in earlier stud- Universitario Parc (http://www.agamenonstudy.com/) and is ies.20 These covariates were performance status Tauli, Sabadell, Spain equipped with multiple tools and filters to guar- (ECOG-PS), histological grade, Lauren histolog- Federico Longo antee data reliability and consistency, as well as ical classification subtype, bone metastases, liver Medical Oncology minimizing missing values; regular telephone and tumor burden, number of metastatic sites, neu- Department, Hospital online monitoring was also conducted. trophil/lymphocyte ratio, chemotherapy regimen, Universitario and HER2 status (IHC 3+ versus IHC 2/FISH+). Ramón y Cajal, Madrid, Spain The study was performed in accordance with Toxicity was classified according to Common Avinash Good Clinical Practice guidelines and the Terminology Criteria for AEs (CTCAE) v4.0. Ramchandani Medical Oncology Declaration of Helsinki. It was approved by Another descriptive variable was the reason stated Department, Research Ethics Committees of all the by the clinician for choosing the chemotherapy Hospital Universitario Autonomous Communities and participating regimen as one of the following alternatives (ina- Insular de Gran hospitals, and was classified as a prospective, bility to swallow, clinical trial, induction therapy Canaria, Las Palmas de Gran postmarketing surveillance study by the Spanish prior to surgery for metastases, tumor regression Canaria, Spain Agency of Medicines and Medical Devices in a symptomatic patient, quality of life, comor- Marta Martín (AEMPS), and was not otherwise involved in it. bidities, clinician’s experience, or department Richard Medical Oncology The list of the Research Ethics Committees protocol). Clinician’s experience refers to the Department, involved in this project is provided in Supplemental oncologist choosing the chemotherapy regimen Hospital Universitario Santa Table S1. All participants still alive at the time of without taking into account considerations Creu i Sant Pau, data collection provided written, signed, informed regarding the patient, the tumor, or the hospital Barcelona, Spain consent. protocol, but instead, their own experience. Alicia Hurtado Medical Oncology Department, The primary objective was to perform an external Hospital Universitario Treatment validation of the results of the ToGA RCT and Fundación Alcorcón, All the subjects received trastuzumab and poly- other trials focused on HER2+ patients.10 Madrid, Spain chemotherapy in first-line treatment. The chemo- Secondary objectives were to assess OS, PFS, Aitor Azkarate Medical Oncology therapeutic regimen, dose intensity, and number ORR, toxicity, subgroup effects, and to ascertain Department, of cycles were chosen at the investigator’s discre- the reason for administering each regimen. Hospital Universitario Son tion. A loading dose of trastuzumab (T) of 8 mg/ Espases, Mallorca, kg of body weight and maintenance dose of 6 mg/ Spain Carolina Hernández kg was administered in the tri-weekly chemother- Statistics Pérez apy regimens. In bi-weekly schemes, trastuzumab Categorical and continuous variables were com- Medical Oncology Department, was administered with both weekly (loading dose pared using Pearson’s Chi-squared tests and two- Hospital of 4 mg/kg and maintenance dose of 2 mg/kg), sample Wilcoxon tests, respectively. The Universitario Nuestra Señora and bi-weekly doses (loading dose of 6 mg/kg and Cochran–Mantel–Haenszel test was used to de the Candelaria, maintenance dose of 4 mg/kg). Trastuzumab was examine the association between categorical vari- Tenerife, Spain administered until progression or unacceptable ables controlling for stratification factors. The Raquel Serrano Medical Oncology toxicity. assumption that the odds ratios (OR) are homo- Department, geneous across several strata was scrutinized by Hospital Universitario Reina means of the Breslow–Day test. Time-to-event Sofía, Córdoba, Variables and objectives was analyzed using multivariable Cox propor- Spain OS was defined as the time between initiation of tional hazards (PH) models, comparing average Javier Gallego Medical Oncology first-line treatment until death from any cause, treatment effect of ToGA regimens with others, Department, censoring subjects who were event-free at their as well as evaluating the effect conditioned on Hospital General Universitario of last follow-up. PFS was defined as the interval of HER2-IHC (tests of interaction). The non-linear Elche, Elche, Spain time between commencement of first-line treat- effect of the neutrophil-to-lymphocyte ratio (con- ment until progression as per RECIST 1.1 crite- tinuous variable) was assessed by means of a ria, demise, or last follow-up. Other variables restricted cubic spline with three knots (a kind of were used for descriptive or auxiliary purposes. smooth function that makes it possible to readily Relative dose intensity was expressed in percent- capture non-linear relations). All analyses were ages and defined as the dose intensity (the amount performed with the R v4.03 software package, of drug per unit of time, expressed as mg/m2 with the rms & ggplot2 libraries.26,27 journals.sagepub.com/home/tam 3 Therapeutic Advances in Medical Oncology 13 Results Trastuzumab-based chemotherapy regimens: dosage and administration Patients Median follow-up in live patients was 62.8 months The AGAMENON-SEOM registry contains (95% CI, 42.8–85.1). At the time of analysis 3018 individuals treated between 2008 and 2020. (November 2020), 82% had discontinued trastu- Within this cohort, 22.3% (n = 675) had HER2+ zumab, with 6.8 months median trastuzumab cancers and 19.6% (n = 594) had received first- treatment duration (95% CI, 6.2–7.5). The line trastuzumab-based chemotherapy. Of the lat- chemotherapy dosage administered, duration, ter, 71% (n = 422) had an IHC 3+ tumor, whereas and criteria for suspension varied depending on 29% (n = 172) had ICH 2+/FISH+. The predic- the scheme. Supplemental Table S5 illustrates tive model of HER2 positivity is shown in the dosages administered. Maintenance therapies Supplemental Table S2. HER2 overexpression with fluoropyrimidine and trastuzumab were was associated with Lauren intestinal or mixed used more frequently following oxaliplatin-based cancers and tumors located in the distal esopha- regimens, especially FOLFOX versus ToGA. gus or gastroesophageal junction. Thus, median exposure to trastuzumab was more prolonged with FOLFOX (10.6 months, 95% CI, Table 1 displays patients’ baseline characteristics. 8.4–20.3) versus CAPOX (7.1 months, 6.1–8.5), Most were intestinal cancers, although the pres- FP (7.0 months, 5.5–7.8), or XP (6.8 months, ence of a diffuse component in one in four and of 5.1–7.9) p < 0.001. Furthermore, participants signet ring cells in 13% is striking. The primary treated with FOLFOX-T received a median of 12 tumor was located in the stomach in 65.3% of the cycles of FU, with tumor progression being the cases (n = 388); in the gastroesophageal junction most common cause for withdrawal. In contrast, in 21.2% (n = 126), and in the esophagus in a median of six cycles of ToGA regimens were 13.4% (n = 80). Of the subjects with progression administered, with the leading reason for discon- at first-line treatment (n = 535), 63% (n = 337) tinuation being “planned treatment completed” received a second line, and 18% (n = 96) received (Supplemental Figure S2). a third line. Safety Selection of chemotherapy regimen Table 2 shows the most common adverse events Different chemotherapy backbones were associ- broken down by treatment type. Most of the tox- ated with trastuzumab (Table 1 and Supplemental icities were mild, although 24.2% (n = 136) of the Table S3), although five of them were adminis- patients were hospitalized for adverse events. Of tered to 87% in the series: capecitabine and cispl- the subjects treated with XP, 30% were admitted atin (XP) in 27% (n = 163), capecitabine and due to toxicity versus around 20% with the oxaliplatin (CAPOX) in 26%, fluorouracil (FU) remaining regimens. and oxaliplatin (FOLFOX) in 14% (n = 86), FU and cisplatin (FP) in 14% (n = 81), and carbopl- Overall, ToGA schemes exhibited the same atin and FU (CarFU) in 5% (n = 30). Alternative hematological toxicity profile as in the original ToGA regimens were prescribed in 59% study, except for more anemia of any grade (XP-T (350/594). The main reasons for choosing a spe- 71%, FP-T 65%). Non-hematological toxicity cific scheme was compliance with local protocol was likewise, comparable, although it depended and the clinician’s experience (Figure 1); other, on the specific regimen. XP-T correlated with less frequent motivations included conversion more grade 3–4 diarrhea compared with FP-T chemotherapy in unresectable disease, preserva- (XP-T 11% versus FP-T 2%), but less stomatitis tion of quality of life, inability to swallow, etc. (XP-T 2% versus FP-T 11%) (Table 2). (Supplemental Figure S1). The characteristics of the subjects treated with the different chemother- Head-to-head comparisons of cisplatin- and apy regimens were comparable except for the oxiplatin-based regimens revealed that the former older age in the CarFU-T-treated group (median associated less grade 3–4 diarrhea (XP-T 4.9%, 72 versus 65 years), worse functional status in the FP-T 2.5% versus CAPOX-T 10.8%, FOLFOX-T FOLFOX-T group (ECOG-PS 2, 25.5 versus 9.3%) and more grade 3–4 thrombosis (XP-T 11.8%), and the higher percentage of diffuse sub- 8.6%, FP-T 8.6% versus CAPOX-T 2.5%, types treated with FP-T (34.6 versus 17.6%) FOLFOX-T 3.5%). Capecitabine-containing (Supplemental Table S4). schemes had less grade 3–4 stomatitis (XP-T 3%, 4 journals.sagepub.com/home/tam P Jimenez-Fonseca, A Carmona-Bayonas et al. Table 1. Baseline characteristics. Table 1. (continued) Variables N (%) 595 (100) Variables N (%) 595 (100) Age, median (range) 65 (22–87)   30–35 93 (15.7) Sex, male 467 (78.6%)   <30 49 (8.2) HER2   No available 55 (9.3)   3+ 422 (71.0) Surgery of primary cancer 141 (23.7)   2+/FISH+ 172 (29.0) Chemotherapy regimen ECOG-PS   XP 163 (27.4) 0 146 (24.5)   CAPOX 158 (26.5) 1 362 (60.9)   FOLFOX 86 (14.4) 2 86 (14.4)   FP 81 (13.6) Charlson index   Carboplatin–fluorouracil 30 (5.0) Cardiovascular chronic disease 61 (10.2)   Anthracycline-based triplet 27 (4.5) Number of chronic comorbidities >2 185 (31.1)   Docetaxel-based regimen 13 (2.1) Tumor stage, locally advanced 23 (3.8)   Modified CAPOX 9 (1.5) unresectable   Other 27 (4.5) Tumor site CAPOX, capecitabine–oxaliplatin; ECOG-PS, Eastern Stomach 388 (65.3) Cooperative Oncology Group Performance Status; FISH, fluorescence in situ hybridization; FOLFOX, fluorouracil– Gastroesophageal junction 126 (21.2) oxaliplatin; FP, fluorouracil–cisplatin; XP, capecitabine– cisplatin. Esophagus 80 (13.4) Lauren subtype   Intestinal 361 (60.8) CAPOX-T 2%, versus FP-T 11% versus FOLFOX-T 9%) and more palmar-plantar eryth-    Diffuse or mixed 122 (20.5) rodysesthesia (CAPOX-T 3%, XP-T 2%, versus   Indeterminate 111 (18.7) FP-T or FOLFOX-T 0%) relative to those based on FU. Toxicity-related mortality was uncom- Signet ring cells 77 (13.0) mon (n = 5, 0.8%) and only occurred in capecit- Metastatic sites abine-containing regimens (CAPOX-T 1.3%, XP-T 1.8%) (Supplemental Figure S3).   Ascites 86 (14.5)   Peritoneal 175 (29.5) Efficacy   Bone 55 (9.3) At approximately 3 months of treatment initia-   Liver 329 (55.4) tion, 437/594 (74%) had measurable and response-evaluable disease. In these cases, the   Lung 143 (24.1) investigator-rated RECIST 1.1 ORR at 3 months   Lymph nodes 307 (51.5) revealed: complete response, 1.1% (n = 5); partial response, 55.8% (n = 244); stable disease, 23.1% Number of metastatic sites, >2 185 (31.1) (n = 101); and tumor progression, 19.9% (n = 87). CEA, median (range) 10 (0–36,000) Individuals with HER2-IHC 3+ cancers had a higher ORR, with common OR 1.97 (95% CI, Albumin 1.25–3.09) (Mantel–Haenszel χ2 = 8.060, df = 1,   Normal (>35 g/dl) 397 (66.8) p = 0.004). This association is homogenous regardless of the scheme used (Breslow–Day (continued) χ2 = 1.062, df = 4, p = 0.900), with no evidence journals.sagepub.com/home/tam 5 Therapeutic Advances in Medical Oncology 13 Figure 1. Reasons for therapy selection. Column percentiles are represented (what led clinicians to select specific regimens in each scenario?). Clinician’s experience refers to the oncologist choosing the chemotherapy regimen without taking into account considerations regarding the patient, the tumor, or the hospital protocol, but instead, their own experience. CAPOX, capecitabine–oxaliplatin; CarboFU, carboplatin–fluorouracil; FOLFOX, fluorouracil–oxaliplatin; FP, fluorouracil– cisplatin; XP, capecitabine–cisplatin. that the ORR differs based on the regimen treated with CAPOX-T or ToGA regimens (χ2 = 3.977, df = 4, p = 0.409). Figure 2 shows a (Figures 3 and 4). PFS data were likewise consist- slope plot with ORR stratified according to HER2 ent. The numbers are small, yet a signal in favor and scheme. of additional benefit with triplets cannot be ruled out (Supplemental Table S6). At the time of analysis, 500 fatal events had been recorded, with median OS of 14.0 months (95% Using the multivariable Cox PH model, HER2- CI, 12.9–15), and 535 progression events, with IHC 2+/FISH+ cancers correlated with worse PFS median PFS of 7.6 (7.0–8.2) months. Figure 3 (HR 1.22, 95% CI, 1.00–1.49) and OS (HR 1.20, illustrates the Kaplan–Meier curves stratified by 0.98–1.47). As for average treatment effects, the chemotherapy scheme. Cox model generates the hypothesis that FOLFOX-T might be superior to ToGA regimens In this clinical practice study, FOLFOX-T with HR for PFS and OS of 0.60 (95% CI, 0.45– reached a median PFS of 9.6 months (7.7–12.9) 0.79) (p < 0.0003) and 0.63 (0.47–0.85) (p = 0.002), and OS of 16.5 months (12.3–24.5), comparable respectively. In a sensitivity analysis, the addition of to the IHC 3+ or IHC 2+/FISH+ subgroup in the variable “reason for treatment choice” (e.g. cli- the reference RCT. Subjects treated with nician’s experience, department protocol, etc.) did CarFU-T achieved the lowest medians with PFS not modify these estimations (see Supplemental of 5.5 months (4.3–7.1) and OS of 8.9 months Figure S4). The inclusion of the secular year as a (5.7–15.7). ToGA regimens displayed the lowest confounding factor did not substantially modify median OS compared with the original RCT: these estimations (datum not shown). 13.7 months (95% CI, 10.6–16.0) for FP-T and 13.9 months (11.4–16.0) for XP-T. Median OS No differences were detected between CAPOX-T was comparable in our study for individuals and ToGA schemes with HR for PFS and demise 6 journals.sagepub.com/home/tam P Jimenez-Fonseca, A Carmona-Bayonas et al. Table 2. Treatment-related toxicity in the most frequent regimens. Toxicity XP (%) CAPOX (%) FOLFOX (%) FP (%) Total Grade 3–4 Total Grade 3–4 Total Grade 3–4 Total Grade 3–4 Anemia 71 9 60 9 81 8 65 7 Neutropenia 57 11 37 22 63 33 60 25 Febrile neutropenia 8 3 8 2 Thrombocytopenia 18 4 31 3 44 5 15 1 Emesis 39 1 32 7 48 2 36 1 Diarrhea 42 11 59 5 46 9 48 2 Stomatitis 32 2 31 3 34 9 57 11 Fatigue 78 5 63 7 78 6 75 4 Heart failure 5 1 2 1 6 1 2 1 Hand-foot syndrome 44 3 43 2 19 0 10 0 Neuropathy 31 4 74 1 74 9 17 5 Alopecia 26 2 8 27 Increased aspartate 7 1 17 1 15 0 9 1 aminotransferase Hyperbilirubinemia 4 1 8 1 6 0 5 0 Nephrotoxicity 9 1 8 1 9 0 9 0 Venous thromboembolic 13 4 11 17 disease* Toxicity-related hospital 30 21 20 20 admission Death due to toxicity 2 1 0 0 *CTCAE v4.0 grades 2–4 are included. CAPOX, capecitabine–oxaliplatin; FOLFOX, fluorouracil–oxaliplatin; FP, fluorouracil–cisplatin; XP, capecitabine–cisplatin. of 0.88 (95% CI, 0.71–1.10) and 0.94 (0.75– although the CI were broad [p(int) = 0.317]. No 1.18), respectively (Figure 4 and Supplemental subgroup effects were noted for hepatic tumor Figure S5). Other Cox model estimates for OS burden, bone metastases, histological grade, or are also displayed in Supplemental Table S7. The for CAPOX-T versus ToGA regimens. non-linear effect of the neutrophil-to-lymphocyte ratio on OS is illustrated in Supplemental Figure S6. Concerning conditional effects, FOLFOX-T Discussion displayed a greater magnitude of effect versus In keeping with the ToGA study, trastuzumab plus ToGA regimens in cancer patients with diffuse cisplatin and capecitabine/fluorouracil is standard (HR 0.37, 95% CI, 0.20–0.69) versus intestinal first-line treatment for HER2+ advanced gastric (HR 0.76, 0.54–1.06) [p(int) = 0.047] subtypes. or gastroesophageal junction adenocarcinoma. The same tendency was seen with respect to Nevertheless, clinicians assiduously use alternative HER2 overexpression intensity, with HR for trastuzumab-associated regimens.8 Ter Veer deaths of 0.47 (95% CI, 0.24–0.92) in IHC2+/ et al.28 compared ToGA regimens against alterna- FISH+ versus 0.69 (0.49–0.96) in ICH3+, tive chemotherapy schemes administered in phase journals.sagepub.com/home/tam 7 Therapeutic Advances in Medical Oncology 13 Figure 2. Slope plots with RECIST categories for HER2 ICH 3+ versus ICH 2+/FISH+. Column percentiles are represented. CAPOX, capecitabine–oxaliplatin; CarboFU, carboplatin–fluorouracil; CR, complete response; FISH, fluorescence in situ hybridization; FOLFOX, fluorouracil–oxaliplatin; FP, fluorouracil–cisplatin; IHC, immunohistochemistry; PD, progression disease; PR, partial response; SD, stable disease; XP, capecitabine–cisplatin. II and observational studies.10 Despite the authors’ different tolerance profile; particularly of note is conclusion that oxaliplatin-based dual-agent the 30% of hospitalizations due to toxicity associ- schemes were preferable to ToGA regimens, the ated with XP-T in our study. methodology was susceptible to confounding bias with consistency among the studies being contra- Second, in this series, CAPOX-T behaves similarly vened and in the presence of problems of external to the ToGA regimens, despite its different safety validity.29 profile (e.g. less thrombosis and stomatitis). As for survival endpoints, our results with CAPOX-T are With this study, we have evaluated the efficacy on a par with those reported by Rivera et al.30 in a and safety of trastuzumab-based chemotherapy Western population. Nevertheless, they are more regimens utilized in the real-world AGAMENON- discreet than those found in both the IHC 3+ or SEOM registry, and arrived at four main conclu- IHC 2+/FISH+ subgroup from the ToGA RCT, sions. First, the ToGA regimens used in the as well as in the two Asian, phase II studies that AGAMENON-SEOM registry achieved worse reported median PFS and OS of approximately 9 survival outcomes compared with the IHC 3+ or and 20 months, respectively11,12 and comprise the IHC 2+/FISH+ subgroup of the original RCT. basis of the meta-analysis by ter Veer.10 The inclusion of non-Asian patients or the greater Consequently, these data similarly fail to replicate frailty of the individuals of this sample (ECOG-PS the Western population of the AGAMENON- 2 in 14% or >2 comorbidities in 31%) might SEOM registry, or point toward a selection bias in account for this discrepancy. Moreover, each spe- favor of patients with better prognoses in both cific scheme (XP-T versus FP-T) has a slightly phase II clinical trials.11,12 8 journals.sagepub.com/home/tam P Jimenez-Fonseca, A Carmona-Bayonas et al. Figure 3. Survival functions stratified by type of treatment. (a) Progression-free survival. (b) Overall survival. CAPOX, capecitabine–oxaliplatin; CarboFU, carboplatin–fluorouracil; CI, confidence interval; FOLFOX, fluorouracil–oxaliplatin; FP, fluorouracil– cisplatin; n, number; OS, overall survival; PFS, progression-free survival, XP, capecitabine–cisplatin. Third, our data support the hypothesis that, in daily particularly relevant in choosing the optimum practice, the average treatment effect of FOLFOX-T backbone of chemotherapy based on ORR and on might be superior to ToGA regimens. This hypoth- the trend in survival endpoints. Thus, should this esis should be tested in a RCT, although it is in line hypothesis that FOLFOX-T might be more effi- with the result reported by Soulareu in a phase II cacious than ToGA regimens be confirmed, it trial with a Western population and incorporated would be especially pertinent in populations who into the meta-analysis by ter Veer.10,13 Tratuzumab benefit less from this strategy, such as diffuse dosing in bi-weekly schedules is different, albeit tumors and those that are HER2+ on IHQ2+/ there are data that suggest that these dosages are FISH+.33 The rest of the literature is compatible capable of reaching the minimum concentrations with this hypothesis. In one pre-planned sub- associated with inhibition of tumor growth.31 One analysis of the ToGA RCT, cisplatin, fluoropy- possible cause for this advantage in the rimidine, and trastuzumab were seen to have a AGAMENON registry is that by using ToGA or greater magnitude of treatment effect in IHC 3+/ CAPOX-T regimens, there was a tendency to FISH+ (HR 0.58, 95% CI, 0.41–0.81) versus assume that the treatment had a pre-defined dura- IHC 2+/FISH+ (HR 0.75, 0.51–1.11) tumors.4 tion as considered for the CF or XP regimens, while The analysis of biomarkers from the GATSBY with FOLFOX-T, FU may have been managed dif- RCT of trastuzumab emtansine in second line ferently, which facilitated maintenance together yielded similar outcomes.34 In parallel, there are with trastuzumab. Optimal treatment duration has data in the literature that speak to a slight superi- not been well defined in advanced gastric can- ority of oxaliplatin dual-agent schemes versus cis- cer.21,32 In the ToGA RCT, individuals without platin in tumors not selected by HER2.6,7 For progression after six cycles discontinued chemo- example, one meta-analysis detected a discreet therapy and maintained trastuzumab until progres- benefit for oxaliplatin (HR for death 0.88, 95% sion.4 However, this strategy may not be the best for CI, 0.78–0.99), which, in addition, was tolerated some patient profiles (e.g. those having a high tumor better in bi-weekly regimens.6 In light of this evi- burden, without toxicity).21 dence, one possible explanation for our results is that the smaller targeted effect of ToGA regimens Beyond the average treatment effects, our study in tumors a priori less sensitive to trastuzumab is suggests that subgroup effects could be offset by the greater cytotoxic spectrum of journals.sagepub.com/home/tam 9 Therapeutic Advances in Medical Oncology 13 non-conventional regimens, such as CarFU-T, and toxicity being recorded in the course of regu- lar care. Despite attempting to control for the most important confounding factors, residual confounding bias cannot be entirely ruled out. Still, there are no data in this study that point toward a selection bias of patients having a better prognosis among those treated with FOLFOX-T, while this regimen tended to be used more in sub- jects having a worse ECOG-PS. Another perti- nent aspect is that HER2-testing was not centrally analyzed, but rather as part of routine practice in each center. The reader should also be aware that the safety profile was assessed solely for descrip- tive purposes, without any attempt to match or adjust for covariates. Finally, despite the fact that the study covers individuals treated over the course of 12 years, the sample does not allow for temporal trends to be analyzed. These trends are generally needed to interpret results in the condi- Figure 4. Cox proportional hazards model for overall survival. tions of ever-changing clinical practice. For the categorical variable “hepatic tumor burden” the contrast is between each level and the rest. The neutrophil-to-lymphocyte ratio variable has been modeled In conclusion, our data emphasize the diversity of by means of a restricted cubic splines, illustrating here the prognostic effect of the criteria and rationale for the use of trastuzumab- increase from 2.2 to 5.1 (25% and 75% percentiles, respectively). CAPOX, capecitabine–oxaliplatin; CarFU, carboplatin–fluorouracil; CI, confidence based chemotherapy regimens in everyday clini- interval; ECOG-PS, Eastern Cooperative Oncology Group performance status; cal practice. This analysis does not support the FISH, fluorescence in situ hybridization; FOLFOX, fluorouracil–oxaliplatin; IHC, conclusion of Ter Veer’s meta-analysis that oxali- immunohistochemistry; NLR, neutrophil/lymphocyte ratio; OS, overall survival; ToGa, platin schemes are more efficacious and safe in ToGa regimens (XP, capecitabine–cisplatin and FP, fluorouracil–cisplatin). general than cisplatin-based regimens, given that in this series of real-world data, the benefit appears to be limited exclusively to FOLFOX-T. oxaliplatin in these groups.4 On the other hand, in Our study generates the hypothesis that this cancers that are more sensitive to trastuzumab scheme may be more efficacious and convenient (intestinal, IHC 3+), trastuzumab’s activity than ToGA regimens, whose efficacy is, in turn, might dilute the lesser efficacy of chemotherapy comparable to CAPOX-T. If confirmed in RCTs, not based on oxaliplatin. This is also in line with the incremental benefit of FOLFOX-T could be various studies that report that trastuzumab is less associated with a pattern of maintaining fluoropy- effective on tumors with heterogenous HER2 rimidine, and is more relevant in tumor subtypes expression, diffuse histopathology, or other a priori less sensitive to anti-HER2 therapy. molecular alterations that provide resistance.35–37 Therefore, these results endorse the future inves- tigation of FOLFOX-T in advanced HER2+ gas- These considerations may be relevant when troesophageal cancer. choosing treatment schemes in future trials with immunotherapy, since, although the evidence is Acknowledgements still limited, preliminary data indicate that oxali- The authors would like to thank the investigators platin would also be more immunogenic than of the AGAMENON study, the Spanish Society other platins, given its greater capacity to induce of Medical Oncology (SEOM), Priscilla Chase immunogenic cell death.38 Finally, CarFU-T Duran for editing the manuscript and Natalia combinations were the least active, even after Cateriano, Gustavo Reporte, Miguel Vaquero, controlling for confounding factors. Data with IRICOM S.A. for supporting the registry triple-agent regimens are still scarce. website. The readers must be mindful that our study has Authors’ contributions several limitations, the most conspicuous one A.C.B., P.J.F., J.G. and A.M.T. developed the being the observational design, use of modified or project, analyzed the data and drafted the 10 journals.sagepub.com/home/tam P Jimenez-Fonseca, A Carmona-Bayonas et al. manuscript. The other authors recruited patients Code availability and provided clinical information, comments, The R Code is available upon request to the and improvements to the manuscript. All authors authors. participated in the interpretation and discussion of data, and the critical review of the manuscript. Supplemental material Supplemental material for this article is available Conflict of interest statement online. The authors declare that there is no conflict of interest. Funding References The authors received no financial support for the 1. Cancer Genome Atlas Research Network. research, authorship, and/or publication of this Comprehensive molecular characterization of article. gastric adenocarcinoma. Nature 2014; 513: 202–209. Ethic approval and consent to participate 2. Deng N, Goh LK, Wang H, et al. A All procedures followed were in accordance with comprehensive survey of genomic alterations the ethical standards of the committee responsi- in gastric cancer reveals systematic patterns of ble for human experimentation (institutional and molecular exclusivity and co-occurrence among national) and with the Helsinki Declaration of distinct therapeutic targets. Gut 2012; 61: 1964 and later versions. Informed consent was 673–684. obtained from all patients before being included 3. Jiménez-Fonseca P, Carmona-Bayonas in the study. This work is original and has not A, Sánchez Lorenzo ML, et al. Prognostic been previously presented elsewhere. significance of performing universal HER2 testing in cases of advanced gastric cancer. Gastric Cancer Ethical statement 2017; 20: 465–474. This study was performed in accordance with the 4. Bang YJ, Van Cutsem E, Feyereislova A, et al. ethical standards of the Declaration of Helsinki Trastuzumab in combination with chemotherapy and its subsequent amendments. This observa- versus chemotherapy alone for treatment of tional, non-interventional trial was approved by HER2-positive advanced gastric or gastro- the Research Ethics Committee of all centers. oesophageal junction cancer (ToGA): a phase 3, open-label, randomised controlled trial. Lancet Informed consent statement 2010; 376: 687–697. Signed informed consent was obtained from all 5. Satoh T, Bang Y-J, Gotovkin EA, et al. Quality alive patients. of life in the trastuzumab for gastric cancer trial. Oncologist 2014; 19: 712–719. Consent for publication 6. Montagnani F, Turrisi G, Marinozzi C, et al. Informed consent and approval by the competent Effectiveness and safety of oxaliplatin compared national authorities includes permission for pub- to cisplatin for advanced, unresectable gastric lication and dissemination of the data. cancer: a systematic review and meta-analysis. Gastric Cancer 2011; 14: 50–55. ORCID iD 7. Enzinger PC, Burtness B, Hollis D, et al. CALGB Nieves Martínez Lago https://orcid.org/ 80403/ECOG 1206: a randomized phase II study 0000-0002-0408-6871 of three standard chemotherapy regimens (ECF, IC, FOLFOX) plus cetuximab in metastatic Data availability esophageal and GE junction cancer. J Clin Oncol All the data generated or analyzed in this study 2010; 28(Suppl. 15): 4006. are included in the manuscript or the supplemen- 8. National Comprehensive Cancer Network. NCCN tary information. clinical practice guidelines for gastric cancer The database is available through a centralized [Internet]. NCCN Guidelines, p. Varsion 2.2020, web platform: http://www.agamenonstudy.com/. https://www.nccn.org/professionals/physician_gls/ pdf/gastric.pdf (accessed 11 March 2019). The details of analyses used in the current study are available from the first author or correspond- 9. Muro K, Van Cutsem E, Narita Y, et al. ing author upon request. Pan-Asian adapted ESMO Clinical Practice journals.sagepub.com/home/tam 11 Therapeutic Advances in Medical Oncology 13 Guidelines for the management of patients of survival in patients with advanced with metastatic gastric cancer: a JSMO–ESMO oesophagogastric adenocarcinoma receiving initiative endorsed by CSCO, KSMO, MOS, first-line chemotherapy: a multicenter prospective SSO and TOS. Ann Oncol 2018; 30: 19–33. study in the era of trastuzumab. Br J Cancer 2017; 116: 1526–1535. 10. ter Veer E, Creemers A, de Waal L, et al. Comparing cytotoxic backbones for first-line 21. Viúdez A, Carmona-Bayonas A, Gallego J, et al. trastuzumab-containing regimens in human Optimal duration of first-line chemotherapy epidermal growth factor receptor 2-positive for advanced gastric cancer: data from the advanced oesophagogastric cancer: a meta- AGAMENON registry. Clin Transl Oncol. Epub analysis. Int J Cancer 2018; 143: 438–448. ahead of print 5 August 2019. DOI: 10.1007/ s12094-019-02183-y. 11. Ryu M-H, Yoo C, Kim JG, et al. Multicenter phase II study of trastuzumab in combination 22. Carmona-Bayonas A, Jiménez-Fonseca P, with capecitabine and oxaliplatin for advanced Echavarria I, et al. Surgery for metastases for gastric cancer. Eur J Cancer 2015; 51: 482–488. esophageal-gastric cancer in the real world: data from the AGAMENON National Registry. Eur J 12. Gong J, Liu T, Fan Q, et al. Optimal regimen Surg Oncol 2018; 44: 1191–1198. of trastuzumab in combination with oxaliplatin/ capecitabine in first-line treatment of HER2- 23. Jiménez Fonseca P, Carmona-Bayonas A, positive advanced gastric cancer (CGOG1001): a Hernández R, et al. Lauren subtypes of advanced multicenter, phase II trial. BMC Cancer 2016; 16: gastric cancer influence survival and response 68. to chemotherapy: real-world data from the AGAMENON National Cancer Registry. Br J 13. Soularue E, Cohen R, Tournigand C, Cancer 2017; 117: 775–782. et al. Efficacy and safety of trastuzumab in combination with oxaliplatin and fluorouracil- 24. Carmona-Bayonas A, Jiménez-Fonseca P, based chemotherapy for patients with HER2- Lorenzo MLS, et al. On the effect of triplet or positive metastatic gastric and gastro-oesophageal doublet chemotherapy in advanced gastric cancer: junction adenocarcinoma patients: a retrospective results from a National Cancer Registry. J Natl study. Bull Cancer 2015; 102: 324–331. Compr Canc Netw 2016; 14: 1379–1388. 14. George SL. Selection bias, phase II trials, and the 25. Carmona-Bayonas A, Jiménez-Fonseca P, FDA accelerated approval process. Oxford: Oxford Custodio A, et al. Anthracycline-based triplets University Press, 2003. do not improve the efficacy of platinum- 15. Kahan BC, Rehal S and Cro S. Risk of selection fluoropyrimidine doublets in first-line treatment bias in randomised trials. Trials 2015; 16: 1–7. of advanced gastric cancer: real-world data from the AGAMEMON National Cancer Registry. 16. Jimenez-Fonseca P, Carmona-Bayonas A, de Gastric Cancer 2017; 21: 96–105. Castro EM, et al. External validity of docetaxel triplet trials in advanced gastric cancer: are 26. Harrell FE Jr. Package ‘rms’. [Internet], p. 229, there patients who still benefit? Gastric Cancer. http://cran.r-project.org/web/packages/rms/index. Epub ahead of print 24 September 2020. DOI: html (2015, accessed 1 January 2020). 10.1007/s10120-020-01116-x. 27. Wickham H. ggplot2: elegant graphics for data 17. Visa L, Jiménez-Fonseca P, Martínez EA, et al. analysis. New York, NY: Springer-Verlag, 2016. Efficacy and safety of chemotherapy in older 28. Ter Veer E, Haj Mohammad N, van Valkenhoef versus non-older patients with advanced gastric G, et al. The efficacy and safety of first-line cancer: a real-world data, non-inferiority analysis. chemotherapy in advanced esophagogastric J Geriatr Oncol 2017; 9: 254–264. cancer: a network meta-analysis. J Natl Cancer 18. Cotes Sanchís A, Gallego J, Hernandez R, et al. Inst 2016; 108: djw166. Second-line treatment in advanced gastric cancer: 29. Jansen JP, Fleurence R, Devine B, et al. data from the Spanish AGAMENON registry. Interpreting indirect treatment comparisons and PLoS One 2020; 15: e0235848. network meta-analysis for health-care decision 19. Carmona-Bayonas A, Jiménez-Fonseca P, making: report of the ISPOR Task Force on Garrido M, et al. Multistate models: accurate indirect treatment comparisons good research and dynamic methods to improve predictions of practices: part 1. Value Health 2011; 14: thrombotic risk in patients with cancer. Thromb 417–428. Haemost 2019; 119: 1849–1859. 30. Rivera F, Romero C, Jimenez-Fonseca P, 20. Custodio A, Carmona-Bayonas A, Jiménez- et al. Phase II study to evaluate the efficacy of Fonseca P, et al. Nomogram-based prediction trastuzumab in combination with capecitabine 12 journals.sagepub.com/home/tam P Jimenez-Fonseca, A Carmona-Bayonas et al. and oxaliplatin in first-line treatment of HER2- previously treated HER2-positive advanced positive advanced gastric cancer: HERXO trial. gastric/gastroesophageal junction cancer. Gastric Cancer Chemother Pharmacol 2019; 83: 1–7. Cancer 2019; 22: 803–816. 31. Felici A, Di Segni S, Contestabile M, et al. Dose 35. Kaito A, Kuwata T, Tokunaga M, et al. HER2 scheduling and pharmacokinetic (PK) study of heterogeneity is a poor prognosticator for HER2- trastuzumab in patients (pts) with HER2/neu positive gastric cancer. World J Clin Cases 2019; overexpressing breast cancer? J Clin Oncol 2007; 7: 1964. 25(Suppl. 18): 1109. 36. Sanchez-Vega F, Hechtman JF, Castel P, et al. 32. Li W, Zhao X, Wang H, et al. Maintenance EGFR and MET amplifications determine treatment of Uracil and Tegafur (UFT) in responders response to HER2 inhibition in ERBB2-amplified following first-line fluorouracil-based chemotherapy esophagogastric cancer. Cancer Discov 2019; 9: in metastatic gastric cancer: a randomized phase II 199–209. study. Oncotarget 2017; 8: 37826. 37. Xu C, Liu Y, Jiang D, et al. Poor efficacy 33. Reichelt U, Duesedau P, Tsourlakis MC, et al. response to trastuzumab therapy in advanced Frequent homogeneous HER-2 amplification in gastric cancer with homogeneous HER2 positive primary and metastatic adenocarcinoma of the and non-intestinal type. Oncotarget 2017; 8: esophagus. Mod Pathol 2007; 20: 120–129. 33185. Visit SAGE journals online 34. Shah MA, Kang Y-K, Thuss-Patience PC, 38. Rébé C, Demontoux L, Pilot T, et al. Platinum journals.sagepub.com/ home/tam et al. Biomarker analysis of the GATSBY study derivatives effects on anticancer immune of trastuzumab emtansine versus a taxane in response. Biomolecules 2020; 10: 13. SAGE journals journals.sagepub.com/home/tam 13