Article Text

Concise report
Comparative effectiveness of switching to alternative tumour necrosis factor (TNF) antagonists versus switching to rituximab in patients with rheumatoid arthritis who failed previous TNF antagonists: the MIRAR Study
  1. Juan J Gomez-Reino1,
  2. Jose Ramon Maneiro1,
  3. Jorge Ruiz2,
  4. Rosa Roselló3,
  5. Raimon Sanmarti4,
  6. Ana Belen Romero5,
  7. on behalf of the MIRAR Study Group
  1. 1Department of Rheumatology, Hospital Clínico Universitario, Universidad de Santiago de Compostela, Santiago, Spain
  2. 2MIXESTAT, SL, Barcelona, Spain
  3. 3Department of Rheumatology, Hospital San Jorge, Huesca, Spain
  4. 4Department of Rheumatology, Hospital Clínic i Provincial, Barcelona, Spain
  5. 5Roche Laboratories, Madrid, Spain
  1. Correspondence to Dr Juan J Gomez-Reino, Department of Rheumatology Unit, Hospital Clinico Universitario, Santiago de Compostela, Santiago, Spain; juan.jesus.gomez-reino.carnota{at}


Objective To compare the effectiveness of switching to rituximab (RTX) with switching to alternative tumour necrosis factor (TNF) antagonists in patients with rheumatoid arthritis (RA) failing on TNF antagonists.

Methods A multicentre prospective 3-year observational study was performed in patients with RA treated with RTX or an alternative TNF antagonist. The baseline 28-joint disease activity score (DAS28) and Health Assessment Questionnaire (HAQ) score were compared with 6, 9 and 12 month values, adjusting for propensity score quintiles. Propensity scores were estimated for each patient using logistic regression with treatment as the dependent variable and baseline prior number of TNFs >1, years from diagnosis >5, extra-articular manifestations, previous toxicity, use of ≥2 disease-modifying antirheumatic drugs, age and sex as independent variables.

Results 1124 patients were treated with either RTX (n=591, 52.6%) or alternative TNF antagonists (n=533, 47.4%). RTX-treated patients had longer disease duration (p=0.0001), larger numbers of previous TNF antagonists (p<0.0001) and tender and swollen joints (p<0.0001). There was no significant difference in the reduction in DAS28 at 6, 9 and 12 months between RTX-treated patients and those treated with TNF antagonists. However, the reduction in DAS28 was significantly different between RTX-treated patients and adalimumab/infliximab-treated patients (p=0.001 and p=0.05, respectively). There was a marginally significant difference at any time period in the proportion of patients achieving an improvement in the HAQ score of >0.22 (p=0.06).

Conclusions Optimal treatment for patients with RA failing on treatment with TNF antagonists may include RTX. This study suggests that the improvement in DAS28 is larger in patients treated with RTX than in those treated with monoclonal anti-TNF agents.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Initial data from observational studies show that switching to the B cell depleting monoclonal antibody rituximab (RTX) may be as effective as, or more effective than, switching to an alternative tumour necrosis factor (TNF) antagonist after initial TNF failure,1 ,2 and data about RTX use in general and TNF class switching are lacking. Moreover, the impact of the order of administration of the TNF antagonists (ie, second or third TNF antagonist) as well as the reason for discontinuation of the prior TNF antagonists on comparative efficacy with RTX is still unsettled.3 ,4 The aim of this study was to determine if patients with rheumatoid arthritis (RA) who fail treatment with TNF antagonists have a better clinical outcome when switching to an alternative TNF antagonist than when switching to RTX, and to try to identify variables that predict the response.


The MIRAR study is a multicentre prospective 3-year observational study of a cohort of patients with RA who received either RTX or an alternative TNF antagonist after failing treatment with at least one TNF antagonist in routine clinical practice. The study was initiated in 2006 in 100 centres across Spain and is reported in accordance with the Preliminary Core Set of Domains and Reporting Requirements for Longitudinal Observational Studies in Rheumatology proposed by Wolfe et al.5

Assessments were made at baseline and for up to 12 months. Time from baseline to follow-up measurements were computed at intervals of 4.5–7.5 months, 7.5–10.5 months and 10.5–12 months and are identified as 6, 9 and 12 months in tables and figures. The erythrocyte sedimentation rate (ESR), swollen joint count and tender joint count based on 28 joints (DAS28), response assessment using European League Against Rheumatism (EULAR) criteria and Health Assessment Questionnaire (HAQ) score were recorded.

Changes in DAS28 and HAQ when switching to RTX versus switching to an alternative TNF antagonist (adalimumab, etanercept and adalimumab) were the main outcomes analysed.

The study was conducted according to good clinical practice applicable to epidemiological studies, which ensures that the design, implementation and communication of data are reliable and that patients’ rights, integrity and data confidentiality are protected. The study was approved by the ethics committee of the Hospital Clinic i Universitary, Barcelona (Spain) and all patients gave their written informed consent.


Descriptive analysis consisted of summary statistics, mean and standard deviation or median and range for continuous variables, and counts and percentages for categorical variables. Propensity scores (PS) were used for comparing baseline characteristics to account for non-random assignment to receive RTX or a TNF antagonist. PS were estimated for each patient using logistic regression with treatment as the dependent variable and the following baseline characteristics as independent variables: prior number of TNF agents >1, years from diagnosis >5, extra-articular manifestations, previous toxicity, use of ≥2 disease-modifying antirheumatic drugs (DMARDs), age and sex.6 ,7 Patients were stratified by quintiles according to their PS, and continuous variables were compared using two-way analysis of variance including PS quintile and treatment as main effects. For categorical variables, controlling for quintile was performed using Cochran-Mantel-Haenszel general association statistic. The quality of the PS adjustment for the stratification approach was evaluated by: (1) computing standardised differences for the unadjusted sample averaged across PS and within each PS stratum, values of standardised difference <0.1 indicating an acceptable balance between treatments; (2) fitting two-way ANOVA (for continuous covariates) or logistic regression models (for binary covariates) with each covariate as the dependent variable and treatment and PS strata and interaction of treatment with PS strata as independent variables to detect possible treatment effect and interaction; and (3) plotting box plots of the PS distributions for each treatment group within each PS stratum.

Changes in DAS28 and HAQ from baseline to 6, 9 and 12 months in the RTX and TNF antagonist groups were compared in the same way, adjusting for PS quintile. Longitudinal change was evaluated, fitting a random coefficient model including DAS28 as a dependent variable and treatment and PS quintile as the main effects. As exploratory analyses, changes in DAS28 were also compared between RTX versus etanercept and RTX versus infliximab/adalimumab. This was done because secondary loss of efficacy could differ with etanercept and anti-TNF monoclonal antibody treatment. Changes in HAQ were compared using the median two-sample test or median one-way analysis. Analyses of EULAR response (good or moderate vs non-response, good vs moderate or no response and good vs no response) and the proportion of patients reaching a HAQ score >0.22 (PS quintiles) were compared using Cochran-Mantel-Haenszel general association statistics.


Baseline characteristics

A total of 1124 patients with RA discontinued at least one TNF antagonist and subsequently received either RTX (n=591, 52.6%) or alternative TNF antagonists (n=533, 47.4%). The two groups were balanced for gender, age and rheumatoid factor ≥60. The RTX group had more severe disease. When comparing the two groups (table 1), significant differences were found in starting treatment with biological agents more than 5 years since the diagnosis of RA (p<0.0001), extra-articular manifestations (p<0.0001), prior number of TNF antagonists failed (p<0.0001), use of ≥2 previous DMARDs (p=0.0028), previous toxicity with DMARDs (p=0.0005) and baseline DAS28 (p<0.0001). Evaluation of balance produced by the propensity model is shown in table 2 stratified by quintiles. All of the averaged standardised differences strata were <0.1. Within PS strata, however, some of differences were >0.1. Within strata box plots (not reported in the paper) showed general agreement and the assessment of treatment by strata interaction in the two-way models was not significant (not reported in the paper).

Table 1

Summary statistics for baseline characteristics before propensity score adjustment

Table 2

Standardised differences before and after adjustment within each propensity score stratum


There was no significant difference in the reduction in DAS28 at 6, 9 and 12 months between RTX-treated patients and those treated with TNF antagonists. However, the reduction in DAS28 was significantly different when comparing RTX-treated patients with adalimumab/infliximab-treated patients pooled together (table 3; p=0.001 and p=0.05, respectively; online supplementary figure). Comparison of RTX with etanercept did not show a significant difference (table 3) at any time period.

Table 3

Changes in disease activity score (DAS) with rituximab versus etanercept or adalimumab/infliximab after propensity score stratification in switching to rituximab compared with switching to alternative tumour necrosis factor antagonists

A significant difference in longitudinal change in DAS28 found when comparing patients treated with RTX with those treated with TNF antagonists disappeared after adjustment for PS (unadjusted p=0.01; PS-adjusted p=0.39). At 6 months there was a significant difference in the proportions of patients achieving good, moderate or no response according to EULAR (table 4; p=0.025). There was a marginally significant difference at each time period in the proportion of patients achieving a HAQ improvement of >0.22 (p=0.06).

Table 4

Proportion of patients responding according to EULAR classification to switching to rituximab compared with switching to alternative TNF antagonists after propensity score stratification


Our results show that, in patients with RA who fail on treatment with a TNF antagonist, switching to RTX is perhaps more efficacious than switching to an alternative TNF antagonist. This is driven apparently by a worse response to switching to adalimumab/infliximab than to etanercept.

Lack of randomisation, patient selection bias and the absence of a washout period are limitations of our study.8 Moreover, observational studies may differ systematically with regard to variables related to exposure or outcome. In this study the RTX-treated group was generally older, had a greater proportion of patients with longer term disease and intolerance to prior treatment and had a greater prior use of DMARDs. This imbalance in baseline data is likely to bias the overall results. However, adjustment by stratification of PS quintiles with regard to known conditional exposure prognostic factors, as we did in this study, allows unbiased estimates and tests of exposure effect.6 ,9 Another potential bias is the incompleteness of data. In the random coefficient model used in our analysis of longitudinal changes, adjustments are made on the basis that missing data are randomly distributed. A sensitivity analysis for data not being randomly distributed could have been done as an alternative analysis. The results should therefore be interpreted with caution. In spite of all these limitations, observational studies provide insights into how used treatments perform in clinical practice.

At 6 months follow-up RTX provided a superior response in a number of outcome measures compared with switching to an alternative TNF antagonist. A significant improvement in DAS up to 9 months was reported in patients treated with RTX compared with patients switched to an alternative TNF antagonist (including etanercept). In the study by Finckh et al around 75% of patients had been switched to a monoclonal anti-TNF antibody whereas in our study this occurred in half of the population. The size of the response of switching to an alternative monoclonal antibody in the analysis of efficacy of TNF antagonists as a whole was smaller in our study, and thus the difference in the change in DAS in comparison with RTX was also less. Nevertheless, the response to RTX treatment according to EULAR was significantly better than the response to alternative TNF antagonists. Interestingly, at 9 months no significant difference was seen in the change in DAS in the RTX and TNF antagonist treatment groups. This could reflect how RTX is used in clinical practice (ie, retreatment when worsening). In fact, this blunting of response to RTX disappears at 12 months when the response is again better than in the TNF antagonist group, perhaps due to the response to retreatment. A somewhat similar picture has been reported by others.1 ,2

In summary, optimal treatment for patients with RA failing on TNF antagonists may include RTX. Our work further suggests that the improvement in DAS28 is larger in patients treated with RTX than in those treated with monoclonal anti-TNF antibodies.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • MIRAR Study Group Alicia García, Juan Povedano, Paula Cejas (Hospital Virgen del Rocío, Sevilla); Federico Navarro, José Alejandro Ariza Ariza, Silvia Ricca (Hospital Virgen Macarena, Sevilla); Juan Sánchez Bursón (Hospital Valme, Sevilla, Spain); Wenceslao Herranz Mediano (Hospital Ntra Sra de la Merced de Osuna, Sevilla); Manuel A. Guzmán Ubeda, Purificación Salas Blanca (Hospital Virgen de las Nieves, Granada); Enrique Raya, Pilar Morales Garrido, Amalia Rueda (Hospital San Cecilio, Granada); Manuel Romero (Hospital General de Jaén, Jaén); Antonio Fernández, Ángeles Belmonte (Hospital Carlos Haya, Málaga); Rosa García (Hospital Virgen Victoria, Málaga); Adela Ruíz (Hospital Reina Sofía, Córdoba); José J. Pérez Venegas (Puerta del Mar Cádiz); José Salberii, Raúl Menor, Yolanda Grandal, Manuel Paez-Camin, Mª del Mar Ruiz (Hospital Jerez de la Frontera, Cádiz); Dolores Mendoza, Manuel Riesco, Celia Fernández (Hospital Juan Ramón Jiménez, Huelva); Concepción Delgado (Hospital Clínico Zaragoza, Zaragoza); Rosa Roselló (Hospital San Jorge, Huesca); Dr. Pina, Dolores Fabregas (Hospital de Barbastro, Huesca); Mª Loreto Horcada (Hospital Obispo Polanco, Teruel); Juan Carlos Torre Alonso (H. Naranco, Asturias); Francisco Javier Ballina García, Mercedes Algeri López, Rubén Queiro Silva, José Luis Riertra Noriega (Hospital Central de Asturias, Asturias); Luis Arboleya (Hospital San Agustín, Asturias); Tomás Tinturé (Hospital Cabueñes, Asturias); Luis Espadaler Poch, Jordi Fiter Areste (Hospital Son Dureta, Mallorca); Inmaculada Ros Vilamajó, Antonio Juan Mai (Hospital Son Llatzer, Mallorca); Teresa Font (Hospital Comarcal d'Inca, Mallorca); Miguel Pericas Alemany (Hospital de Manacor, Mallorca); Ana Urruticoechea Arana (Hospital Can Misses, Ibiza); Sagrario Bustabad, Ivan Ferraz (HCU Canarias, Tenerife); Alberto Cantabrana (Hospital Candelaria, Tenerife); Daniel Batista, J Angel Hdez (Hospital Insular, Las Palmas); Juan Carlos Quevedo, Íñigo Rúa- Figueroa (Hospital Dr. Negrín, Las Palmas); Jaime Calvo, Elena Arrecoechea, Teresa Jimeno (Hospital Sierrallana Torrelavega, Cantabria); Raimon Sanmarti, M. Victoria Hernández (Hospital Clinic, Barcelona); Carme Moragues, Ingrid Möller (Clínica Platón, Barcelona); Nishishinya Mª Batina, Díaz Torné, Vicente Torrente Segarra, Asunción Acosta (Hospital Sant Pau, Barcelona); Mª Bonet (Hospital de Vilafranca, Barcelona); Eduardo Kanterewicz, José Rodriguez (Hospital de Vic, Barcelona);Ramón Figuls (Creu Roja Barcelona, Barcelona); Marta Larrosa Padró, Antonio Gómez (Hospital Parc Tauli de Sabadell, Barcelona); José Valverde García, Elena Sirvent, Delia Reina Sanz (Hospital Bellvitge); Jose Miguel Ruiz, Sergio Ros (Hospital Viladecans, Barcelona); Jordi del Blanco (Hospital de Calella, Barcelona); Joan Maymó, Mapi Lisbona, Carolina Pérez, Miriam Almirall, Núria Aizpunua Perez, Silvia Sánchez (Hospital del Mar, Barcelona); Alejandro Olivé, Elisa Garcia (Hospital German Trias i Pujol, Barcelona); Georgina Salvador, Elena Riera (Hospital Mutua Terrassa, Barcelona); Dra. Marsal (Hospital Vall d'Hebrón, Barcelona); Maria Isabel Rotés, Alba Erra, Estefania Moreno (Hospital Sant Rafael, Barcelona); Teresa Claveguera, Ramón Valls (Hospital Palamós, Girona); Miquel Sala, Marta Valls (Hospital Figueres, Girona); Lluis Roselló (Hospital Sta. María, Lleida); Sergi Ordoñez Palau (Hospital Arnau de Vilanova, Lleida); Silvia Paredes (Hospital St Joan Reus, Tarragona); Josep Pujol, Rosa Morlà (Hospital St Pau i Sta. Tecla, Tarragona); Ramon Fontova Garrofí, Mª José Poveda Elias (Hospital Joan XXIII, Tarragona); X. Arasa (Hospital Verge de la Cinta, Tarragona); Alfons Lorenzo Querol, Mª José Poveda Elias (Pius Hospital de Valls, Tarragona); Enrique Judez, Gloria Garcia (Complejo Hosp. Albacete, Albacete); Paula Virginia (Complejo Hosp. Ciudad Real); Julio Vázquez (Hospital Virgen de la Luz, Cuenca); Pilar Muñoz (Hospital Guadalajara, Guadalajara); Juana Sampedro Álvarez, Angel M García Aparicio, Azucena Hernández Sanz, Jose Rey Rey (Hospital de Toledo, Toledo); Montserrat Corteguera (Hospital Ntra. Sra. Sonsoles, Avila); Manuel Martín Martínez (Hospital de León, Leon); Lucia Pantoja (Hospital del Bierzo, Leon); Dr. Montilla (Hospital Virgen de la Vega, Salamanca); Purificación Ruiz (Hospital Insalud Soria, Soria); JM Martín Santos (Hospital Río Ortega, Valladolid); Jose Luis Alvarez Vega, Jose Maria Salazar (Hospital Infanta Cristina, Badajoz); Eugenio Chamizo, Juan Jose Aznar, Noemí Garrido (Hospital Mérida, Badajoz); Antonio Cardenal Escarcena, Antonia Ferreira Conejo (Hospital Don Benito, Badajoz); Raúl Veroz González (Hospital Llerena, Badajoz); José García Toron (Hospital San Pedro Alcántara, Cáceres); Miguel Ángel Abad, María Torresano Andrés (Hospital Virgen del Puerto, Cáceres); Mercedes Freire, Jenaro Graña (Hospital Juan Canalejo, A Coruña); Marina Rodriguez López, Mª Ángeles Hernández (Complejo Hospitalario Arquitecto Marcide- Novoa Santos, A Coruña); Jorge Blanco Rodríguez, Juan José Díaz Garel (Complejo Hospitalario Universitario de Santiago, A Coruña); Manuel Rodríguez Fernández, Luis Fernández Domínguez, Alfredo Willisch Domínguez (Hospital Cristal Piñor, Ourense); Juan Cruz Martinez (Hospital provincial de Pontevedra, Pontevedra); Rubén Ouviña, Iñigo Hernández (Hospital Mexoeiro, Pontevedra); Ceferino Barbazán (Hospital Xeral Cíes; Pontevedra); José Ángel Cabezas (Hospital San Millán y San Pedro, La Rioja); Julio Gracia Estevez (Hospital general de la Rioja, La Rioja); Dra. Mateo, Beatriz Joven, María Galindo, Luis Morillas, Pilar Fernández Dapica, Laura González Hombrados, Padrino Herrera, Javier García (Hospital 12 de Octubre, Madrid); María Alcalde, Esperanza Naredo, Paz Collado, Dr. Crespo (Hospital Severo Ochoa, Madrid); Luis Carreño, Dr. Gonzalez, Francisco Javier López, Marilyn Salvat, Indalecio Monteagudo (Hospital Gregorio Marañón); Esther Vicente Rabaneda, Esther Coledaro (Hospital Princesa, Madrid); Dra. Sánchez Atrio (Hospital Príncipe de Asturias, Madrid); Tatiana Cobo, Martina Steiner (Hospital La Paz, Madrid); Paloma de Abreu, Oscar Illera (Hospital Ramón y Cajal, Madrid); Ana Castilla Plaza (Hospital Carlos III, Madrid); Roberto Mígueles Sánchez, Mª Cruz Fdez Espartero, Virginia Villaverde García, Mercedes Morcillo Valle, Jacqueline Usón Jaefer (Hospital Móstoles, Madrid); Dr. Prada, Dr. Mulero (Hospital Puerta de Hierro, Madrid); Pedro Zarco (Fundación hospital de Alcorcón, Madrid); María Luisa González (Hospital Comarcal El Escorial, Madrid); Ángel Aragón (Hospital de Getafe, Madrid); Carlos Rodríguez Escalera (Hospital Comarcal de Melilla, Melilla); Francisco García Villalba, Manuel Moreno Ramos, Juan Moreno Morales (Hospital Santa María del Rosell, Murcia);Carmelo Tornero, Encarnación Saiz Cuenca (Hospital Morales Moseguer, Murcia); Carlos Marras, Nuria Lozano (Hospital Virgen de la Arrixaca, Murcia);Emilia Aznar (Hospital Reina Sofía, Navarra); Eduardo Loza Cortina, Concepción Fito, Natividad del Val, Rosario Ibañez, Ricardo Gutierrez (Hospital de Navarra, Navarra); Juan Carlos Vesga (Hospital de Txagorritxu Vitoria, Álava); Alberto Alonso Ruiz (Hospital de Cruces, Vizcaya); José Miguel Aramburu Albizuri, M. Luz García, Eva Galinder, Eduardo Ucar, Manuel Gorordo, Arantza Pacho, Iñaki Torre, Francisco García, Olaia Fernandez (Hospital de Basurto, Vizcaya); Flores Torre, Aintzane Ruiz Diez (Hospital de Galdákano, Vizcaya); Joaquín Belzunegui (Hospital de Donostia, Guipuzcoa); Paloma Vela (Hospital General Alicante, Alicante); Dra. Jovaní, Cristina Fernández (Hospital General de Elda, Alicante); Miguel Belmonte (Hospital General de Castellón, Castellón); José A. González, Francisco Navarro Blasco, Juan Tovar Beltran, José Raúl Noguera (Hospital General Elche, Alicante); Mauricio Minguez (Hospital Clínico San Juan, Alicante); Francisco Pérez (Hospital de Requena, Valencia); Pascual Muñoz (Hospital de Xativa, Valencia); Dra. Martínez (Hospital Alzira, Valencia); Dra. Alcañiz (Hospital Dr. Peset, Valencia); Antonio Gracia Pérez (Hospital Sagunto, Valencia); Juan Antonio Castellano, Antonio Pérez (Hospital Arnau de Vilanova, Valencia); Isabel González -Cruz (Hospital General de Valencia, Valencia); Hospital Marqués de Valdecilla, Cantabria; Hospital Miguel Servet, Zaragoza.

  • Contributors JJG-R: conception and study design, interpretation of data, drafting the article, revising it critically for important intellectual content. JRM: drafting the article. JR, RR, RS, ABR and JRM: data collection and interpretation of data. All authors gave final approval of the version to be published.

  • Funding The study was supported by Roche, Spain.

  • Competing interests JJG-R is on the Advisory Boards of BMS, Pfizer, Roche, Schering-Plough and UCB SA; has received lecture fees from, BMS, Roche, Schering-Plough and Wyeth; and has received research grants from Roche and Schering-Plough. All other authors have declared no conflicts of interest.

  • Patient consent Not obtained.

  • Ethics approval Ethics approval was obtained from the ethics committee of the Hospital Clinic i Universitary, Barcelona (Spain).

  • Provenance and peer review Not commissioned; externally peer reviewed.