Rivaroxaban: can we trust the evidence?

We read with interest the recent BMJ editorial “Rivaroxaban:Can we trust the evidence”.1 which questions the conclusions from the ROCKET AF clinical trial on Rivaroxaban 2. Cohen raises questions about the accuracy of the INRs used for monitoring patients in the warfarin control arm and suggests that they may have compromised the validity of the published conclusions. The ROCKET investigators have defended their conclusions after re assessing the data in light of these concerns 3. The device used for INR testing in the warfarin control arm was not identified in the original publication but subsequently has been revealed as the INRatio Point of Care (POC) device. This device was the subject of a safety recall notice by FDA in 2014 due to “clinically significantly lower” readings compared to laboratory values, with the fault going back to 2002 (before the Rocket AF trial started). The BMJ editorial points out that a falsely low INR reading could mean that patients in the control arm may have had their warfarin dose unnecessarily increased leading to a greater risk of bleeding.

To ensure accuracy, POC INR testing should be subjected to the same level of quality control (QC) and quality management systems that laboratory INR testing has to meet. Appropriate and effective External Quality Assessment (EQA) for POC INR testing devices is possible 4 and there has been a POC INR EQA programme available in the UK for 20 years. In the UK NEQAS for Blood Coagulation EQA programmes we provide suitable material for several different devices and have over 4500 users enrolled in the programmes (January 2016). The POC device used in the warfarin control arm of the ROCKET AF study 2 is one which we evaluated to assess our EQA material suitability in 2010 but found we were unable to provide any suitable EQA material for this technology. To our knowledge no other programme provides EQA for this device. The lack of EQA and internal quality control (IQC) makes it difficult to assess the reliability of INR results obtained with this device.

Laboratory based INR testing in many countries is subject to accreditation, often against international ISO standards (ISO 15189) 5 which require participation in regular EQA as part of the quality management system. Currently the vast majority of POC INR testing in many countries is not subject to such accreditation and not all POC sites participate in EQA even when this is available. Although such participation for POC sites is not currently mandated by regulation, we believe it should be obligatory. Non-participation despite availability could have serious medico-legal implications in the event of untoward incidents. If the use of INR methods in the laboratory or in point of care was restricted to those for which satisfactory participation in EQA can be demonstrated, the issues of clinically significant problems in INR testing such as those which have led to the questioning of the ROCKET AF study could be more easily avoided.

POC INR testing has been in use for over two decades and in our view is safe, providing continuous QC assessment is in place in order to ensure that the INR results produced are valid and accurate 6.There are several published national and international guidelines stating the essential requirement for QC to ensure reliable results 7-9 when using POC INR devices. It would be inappropriate to view all POC devices for INR testing to be inferior to the laboratory measurements as this is not usually the case. POC devices can be accurate and precise giving reliable results in a very quick turnaround time. The key is to ensure the accuracy and precision of all INR results, whether derived from a POC device or from a laboratory method, with the use of both internal control and external quality assessment.

References
1.Cohen D. Can we trust the evidence? BMJ 2016,352,179-180

2. Patel MR, Mahaffey KW, Garg J, Pan G, Singer DE, Hacke W, Breithardt G, Halperin JL, Hankey GJ, Piccini JP, Becker RC, Nessel CC, Paolini JF, Berkowitz SD, Fox KAA, Califf RM, and the ROCKET AF Steering Committee, for the ROCKET AF Investigators* N Eng J Med 2011:365; 883-891

3.Patel MR, Hellkamp AS, Fox KAA, and the ROCKET AF executive committee and investigators. Point–of–Care Warfarin monitoring in the ROCKET AF Trial. N Eng J Med 2016:374-378

4 Kitchen DP, Kitchen S, Jennings I, et al. Point of Care INR testing devices: performance of the Roche CoaguChek XS and XS Plus in the UK NEQAS BC external quality assessment programme for healthcare professionals: four years’ experience. J Clin Pathol 2012;65(12):1119–1123

5 The British Standards : Medical laboratories- Requirements for quality and competence ISO 15189

6. Kitchen S, Kitchen DP , Jennings I, Woods TAL, Walker ID , Preston FE. Point of Care INRs : UK NEQAS experiences demonstrates necessity for proficiency testing of three different monitors. Thromb Haemost 2006 : 96; 590-6.

7.CLSI Point –of –Care monitoring of anticoagulation therapy: Approved guideline. CLSI document POCT14-A Wayne PA:Clinical and laboratory standards institute 2004

8. Keeling D, Baglin T, Tait C, et al; British Committee for Standards in
Haematology. Guidelines on oral anticoagulation with warfarin -fourth edition. Br J Haematol 2011;154(3):311–324

9. Jennings I, Kitchen D, Keeling D, et al; BCSH Committee. Patient self-testing and self-management of oral anticoagulation with vitamin K antagonists: guidance from the British Committee for Standards in Haematology. Br J Haematol 2014;
167(5):600–607

It is with concern that we read the Editorial in the BMJ (1) regarding the ROCKET-AF trial comparing rivaroxaban (Xarelto®) with a warfarin treated control group. It has been brought forth that the warfarin-managed group had their INR measured on a coagulometer, which has some serious drawbacks. Here we will focus on the point of care coagulometer (POCT) used in the trial; discuss the impact on the results and the requirements for the use of a POCT in trials.

Generally, in order to provide solid evidence in terms of non-inferiority (or even superiority) of rivaroxaban compared with warfarin, the latter needs to have a sufficiently high time within therapeutic target range (TTR) – preferably ≥70%. In the ROCKET-AF trial (2) the TTR in the warfarin group was merely 55%, which reflects a suboptimal quality of oral anticoagulant therapy (3). Using patient self-testing and self-management a higher quality of treatment can be achieved for selected patients. (4) High TTR has been linked to improved efficacy and safety outcomes.(5)

In a previous paper we concluded, that the precision of the POCT coagulometers was generally adequate for clinical use, but their performance in terms of accuracy has to be viewed in the context of the inherent inaccuracies in INR measurements. The overall accuracy of POCT coagulometers seems, in this respect, to be generally acceptable for everyday clinical practice and they may be used in a clinical setting.(6) We also showed that the HemoSense (INRatio®) had not been investigated in any high quality studies and the precision and accuracy of that specific coagulometer was lower compared to some of the other coagulometers on the market. Therefore, one may wonder from an analytical perspective why the investigators of the ROCKET-AF study choose to use HemoSense (INRatio®) in a clinical trial?

A potential underestimation of the INR value compared with a laboratory standard is the concern. This could potentially lead to a higher risk of bleeding in the warfarin group and subsequently misleadingly favour rivaroxaban over warfarin in a direct comparison on safety outcomes. The authors of the ROCKET-AF trial have responded that INR values were merely low in patients with certain conditions in terms of e.g. chronic inflammation, acute inflammation, or hematocrit levels out of range. Furthermore, it is advocated, that this did not influence the conclusions of the study regarding safety endpoints.(7) It should be emphasized that a statistical insignificant result may still be regarded as a clinical relevant result. Additionally, it can be seen that the previous statistically significant and clinically relevant reduction in intracerebral haemorrhage for the rivaroxaban arm was not observed in the group with the conditions mentioned above. For the outcome of major bleeding the subgroup of patients with none of the conditions favoured treatment with rivaroxaban compared with warfarin, hazard ratio 0.87 (95% CI, 0.70-1.08); while patients with any of the conditioned displayed a hazard ratio of 1.18 (95% CI, 0.98-1.42).7 These results are counterintuitive with the hypothesis that patients with the mentioned conditions could have received a higher dose of warfarin resulting in an increased risk of bleeding. It could also indicate that:

1) This subgroup is not the patients causing problems in relation to INR measurements,
2) The analytical error does not affect the over all conclusion.

However, the overall statement that the results of the coagulometer (and conditions) had no influence on the outcome is somewhat speculative, and further enhances the doubt on the conclusions from the trial that has already been raised.(1)

It has been argued that a quality control of these coagulometers is necessary, which can be performed using several different setups.(8) Hereby, flaws in INR measurements than can be attributed to analytical variations would probably be detected, but these quality controls would not predict flaws attributable to the biological deviations in the individual patient as those described above.

In a clinical trial there must be transparency regarding which type of coagulometer is used, and the quality of INR measurement should be high and precise. This could have been fortified in the ROCKET-AF trial by using a more validated coagulometer and quality control of the INR values, e.g. by using INR measured in a certified laboratory.

We recommend that coagulometers used in a clinical study should be those with the evidence of the best performance in terms of accuracy and precision. Otherwise, the obtained results should be interpreted with caution and may limit the clinical impact in terms of changing treatment strategies.

References
1 D Cohen. Rivaroxaban: can we trust the evidence? BMJ 2016;352:i575
2 Patel MR, Mahaffey KW, Garg J, et al. Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med 2011;365:883-91.
3 Christensen TD, Larsen TB, Hasenkam JM. Results of Vitamin K antagonist treatment in patients with atrial fibrillation depends on the type of management. Circulation 2012;126:e45.
4 Nilsson H, Grove EL, Larsen TB, et al. Sex differences in treatment quality of self-managed oral anticoagulant therapy: 6,900 patient-years of follow-up. PLoS One 2014;9:e113627.
5 De Caterina R, Husted S, Wallentin L, et al. Vitamin K antagonists in heart disease: current status and perspectives (Section III). Position paper of the ESC Working Group on Thrombosis--Task Force on Anticoagulants in Heart Disease. Thromb Haemost 2013;110:1087-107.
6 Christensen TD, Larsen TB. Precision and accuracy of point-of-care coagulometers used for self-testing and self-management of oral anticoagulation therapy. Journal of Thrombosis and Haemostasis 2012;10:251-260.
7 Patel MR, Hellkamp AS, Fox KAA, et al. Point-of-Care Warfarin Monitoring in the ROCKET AF Trial. N Engl J Med 2016 Feb 3 [Epub ahead of print].
8 L Poller. Precision and accuracy of CoaguChek S and XS monitors: The need for external quality assessment. Thromb Haemost 2009;101:419–421.

The BMJ must continue its campaign for open access to RCT data, without fear or favour (http://www.bmj.com/content/350/bmj.h3337).

Sadly, in an intensely competitive Pharma world, the companies are unlikely to throw open their most intimate secrets.. but they could confirm how accurately their INR data was lab-validated ??

At least the Rocket-AF triallists have reviewed the effect of the recalled device, thus - http://www.nejm.org/doi/full/10.1056/NEJMc1515842?query=featured_home