Therapeutic monitoring of TNF-alpha inhibitors in Crohn’s disease (LISA-TRACKER ELISA kits, IDKmonitor ELISA kits, and Promonitor ELISA kits)

The Diagnostics Advisory Committee considered the impact of Crohn's disease on a person's life. It heard from a clinical expert on the Committee that the disease most often presents in early adulthood, but can occur at any age. It heard that severe Crohn's disease can have devastating effects on a person's life, such as extreme weight loss, fistulas and abscesses, the need for surgery, and enteral or parenteral nutrition. The Committee heard from a patient expert that even mild or moderate Crohn's disease can have a substantial impact on a person's day‑to‑day quality of life in the form of fatigue, fever, anaemia, diarrhoea and joint pain. It also heard that having Crohn's disease can result in a person needing substantial time off work and can restrict their participation in activities with their family. The Committee concluded that Crohn's disease substantially impacts the quality of life of the person with Crohn's disease and their family.

The Committee considered the complexity of managing Crohn's disease. It heard from a clinical expert that many different factors influence the development and progression of Crohn's disease, including genes and the environment. It also heard that because of these different influences, adapting treatment to suit individual patients can be difficult and there are limited options for treatment, particularly in those with severe disease. The Committee concluded that managing Crohn's disease is extremely complex and that new tests that can help clinical decision‑making could improve management of the condition and improve outcomes for the patient.

The Committee reviewed the evidence available on the clinical and cost effectiveness of using enzyme‑linked immunosorbent assay (ELISA) kits (LISA‑TRACKER, IDKmonitor, and Promonitor) to test levels of tumour necrosis factor (TNF)‑alpha inhibitors and antibodies to TNF‑alpha inhibitors in the following 2 populations:

• people with Crohn's disease whose disease responds to treatment with a TNF‑alpha inhibitor

• people with Crohn's disease whose disease loses response during maintenance treatment with a TNF‑alpha inhibitor.
  
  The Committee noted that no clinical outcome data were available on the 3 index tests (LISA‑TRACKER, IDKmonitor and Promonitor ELISA kits) and therefore the results of the economic model were based on the results from studies of alternative tests (Prometheus ELISA, Prometheus homogeneous mobility shift assay [HMSA]), radioimmunoassay and Leuven in house ELISA).

The Committee considered the test performance of the ELISA kits (LISA‑TRACKER, IDKmonitor and Promonitor) compared with the alternative tests that have direct clinical outcome data (Prometheus ELISA, Prometheus HMSA, radioimmunoassay and Leuven in‑house ELISA). It noted that the evidence base on comparative test performance was very small, which led to great uncertainty in the comparability of the different tests. The Committee heard from clinical experts that most testing in the UK is done in a few centres with each using different test kits or laboratory‑developed methods. The Committee concluded that because of the absence of clinical data, it was uncertain which of the tests would be most clinically useful in both scenarios. It concluded further, that in the absence of robust positive or negative evidence linking the index tests to the alternative tests, the outcomes of the economic model can be applied to the index tests (LISA‑TRACKER, IDKmonitor and Promonitor ELISA kits), and the Committee noted the uncertainty in making this assumption.

The Committee considered the analytical validity of the ELISA kits (LISA‑TRACKER, IDKmonitor and Promonitor). It heard from experts on the Committee that measuring TNF‑alpha inhibitor levels can be problematic if antibodies to TNF‑alpha inhibitors are also present in the sample. It heard further that tests measuring antibodies to TNF‑alpha inhibitors do not distinguish between transient antibodies (antibodies that disappear and reappear over time) and stable antibodies (antibodies that stay at high levels), but that the type of antibody is clinically important and could affect treatment decisions. The Committee noted that some ELISAs for antibodies to TNF‑alpha inhibitors are quantitative and others are semi‑quantitative, and concluded that it was uncertain which would be most clinically useful. The Committee questioned whether the ELISAs would work with biosimilar versions of TNF‑alpha inhibitors. It heard from an expert on the Committee that studies have been done that show the ELISAs do work with biosimilars, but these studies are currently unpublished. The Committee also heard from a clinical expert that there is no formal external quality assurance programme for measuring levels of TNF‑alpha inhibitors and antibodies to TNF‑alpha inhibitors, but that some laboratories take part in sample‑exchange programmes as a form of quality assurance. The Committee concluded that further research into the analytical performance of the ELISAs is needed.

The Committee considered the evidence on the optimal cut‑off thresholds for use with the ELISA kits (LISA‑TRACKER, IDKmonitor and Promonitor). It noted that the information‑for‑use documents for each of the kits do not specify thresholds to guide interpretation of test results. So each laboratory is expected to identify and validate a threshold for use with the tests. The Committee heard from clinical experts that the same thresholds should not be used between different kits, making it difficult to compare the results from different kits. It also heard that thresholds for infliximab levels were better established than thresholds for adalimumab levels, and that thresholds for TNF‑alpha‑inhibitor levels were better established than thresholds for antibodies to TNF‑alpha inhibitors. The Committee noted that a precise threshold was less critical for people whose disease loses response, because the objective of testing was to identify the presence or absence of the TNF‑alpha inhibitor. It noted further that a precise threshold was more important in people whose disease was responding to treatment with a TNF‑alpha inhibitor in whom the objective of testing was to titrate the dose of the TNF‑alpha inhibitor to achieve a trough level in a target range. The Committee therefore concluded that further research is needed to establish clinically meaningful thresholds for each of the ELISAs, and considered that laboratories currently doing these tests should have specialist expertise in immunoassay analysis and should be interpreting the results with caution.

The Committee considered the accuracy of the ELISA kits (LISA‑TRACKER, IDKmonitor and Promonitor) for predicting the clinical state of Crohn's disease. The Committee noted that evidence suggested that the ability of these tests to classify clinical state is poor, which could result in misclassification of clinical state in people with Crohn's disease. The Committee concluded that the uncertainty in the accuracy of these tests would lead to uncertainty in the model, and that further research is needed on the clinical validity of the ELISA kits.

The Committee considered the outcomes used to assess the response of Crohn's disease to treatment with TNF‑alpha inhibitors. It heard from an expert on the Committee that the outcomes often used to assess response include blood serum biomarkers and mucosal healing (through endoscopy). It heard further that the outcome most important to people with Crohn's disease is that they feel better. The Committee noted that the levels of TNF‑alpha inhibitor needed to achieve mucosal healing are higher than the levels of TNF‑alpha inhibitor needed for the person to feel better. It concluded that future studies should include patient‑reported outcomes measures.

The Committee considered the assumptions used in the economic models. It noted that 2 different sets of transition probabilities were used to generate 2 sets of base‑case results. The first base case used time‑dependent transition probabilities, whereas the second base case used exponential transition probabilities. The Committee heard from the External Assessment Group (EAG) that the time‑dependent transition probabilities best reflect the data from the key studies used to provide inputs for the model. However, the exponential transition probabilities assume that people progress through the model at a constant rate over time, and this better reflects how people move through the care pathway, given the modelling methods that were used. The Committee concluded that the results of the second base case were the most plausible.

The Committee considered the test schedules assessed in the models. It noted that in the base case, people whose disease responded to TNF‑alpha‑inhibitor treatment were tested for TNF‑alpha‑inhibitor levels and antibodies to TNF‑alpha inhibitors every 3 months. The Committee heard from a clinical expert that in UK practice the most likely testing strategy is to test for TNF‑alpha‑inhibitor levels and antibodies to TNF‑alpha inhibitors once a year and on loss of response. The Committee concluded that the most plausible ICER for the responder model was from the 'annual testing' scenario (£126,600 saved per QALY lost for concurrent testing compared with no testing), and the most plausible ICER for the loss of response model was from the 'testing only on loss of response' scenario (£340,900 saved per QALY lost for concurrent testing compared with no testing).

The Committee considered the QALY losses resulting from the economic model that were spread over 10 years. It noted that in the responder model with annual testing, the QALY losses compared with a no‑testing strategy were 0.280 for a reflex‑test strategy and 0.288 for a concurrent‑test strategy. It also noted that in the loss of response model, when testing was done only in people whose disease lost response to a TNF‑alpha inhibitor, the QALY losses compared with a no‑testing strategy were 0.333 for a reflex‑test strategy and 0.351 for a concurrent‑test strategy. The Committee considered that these QALY losses are quite large and unexpected, given the low quality of life experienced by people with Crohn's disease that loses response to a TNF‑alpha inhibitor. The Committee also heard from a clinical expert on the Committee that the quality of life of people with Crohn's disease can be difficult to value, and that utility values used in the model may not fully reflect the quality of life of people with Crohn's disease. The Committee concluded further that research into the quality of life of people with Crohn's disease treated with TNF‑alpha inhibitors would be useful.

The Committee considered the reasons for the QALY losses. It heard from the EAG that one reason was the high proportion (79%) of people in the model in the loss of response health state with TNF‑alpha inhibitor present and antibodies to TNF‑alpha inhibitors absent. This proportion was taken from the study by Steenholdt et al. (2014). A clinical expert on the Committee advised that the proportion of patients in the UK with loss of response, TNF‑alpha inhibitor present and no antibodies to TNF‑alpha inhibitors is much lower than 79%. The Committee also heard from the EAG that these people in the model, with a testing strategy, would stop TNF‑alpha‑inhibitor treatment and have best supportive care, which would eventually include surgery. A no‑testing strategy would result in these patients staying on TNF‑alpha‑inhibitor treatment longer before stopping the TNF‑alpha inhibitor and having best supportive care. The Committee also noted its conclusion that the uncertainty in the accuracy of the ELISA kits for predicting clinical state could lead to misclassifying clinical state in some people with Crohn's disease (section 6.7). It noted further that if clinical state in people with Crohn's disease is being misclassified by the test results this may explain some of the QALY losses seen in the economic model. The Committee concluded that the QALY losses in the models were uncertain and may not reflect clinical practice in the NHS.

The Committee considered the cost savings resulting from the economic model. It noted that the cost savings in the testing strategies compared with the no‑testing strategy were driven by reduced use of TNF‑alpha inhibitor in the testing strategies, particularly by:

• not increasing the dose of TNF‑alpha inhibitor in people whose disease loses response and who have high levels of antibodies to TNF‑alpha inhibitor

• stopping treatment with, or reducing the dose of TNF‑alpha inhibitor in people whose disease is in remission and have undetectable or low trough level of TNF‑alpha inhibitor.
  
  The Committee heard from an expert on the Committee that biosimilars for infliximab and adalimumab are likely to be introduced soon. The Committee noted that biosimilar drugs are cheaper than the original drugs, which would be likely to reduce the cost savings in the model. The Committee concluded that the small evidence base led to uncertainties in the modelling, which resulted in uncertainty in the cost savings.

The Committee considered the probabilistic sensitivity analyses done by the EAG. It noted that the scatterplots for both the loss of response model and the responder model showed considerable uncertainty in the QALY losses, and that there was overlap between the results of the testing strategies and the no‑testing strategy in terms of QALY losses. The Committee also noted that the scatterplot for the responder model showed considerable uncertainty in the cost savings, and there was overlap between the results of the testing strategies and the no‑testing strategy in terms of cost savings. However, the scatterplot for the loss of response model showed slightly less uncertainty in the cost savings, and there was no overlap between the results of the testing strategies and the no‑test strategy in terms of cost savings. The Committee concluded that there was greater uncertainty in the cost savings in people whose disease was responding to TNF‑alpha‑inhibitor treatment compared with people whose disease loses response to TNF‑alpha‑inhibitor treatment.

The Committee considered the current UK use of testing for TNF‑alpha inhibitors and antibodies to TNF‑alpha inhibitors in Crohn's disease. It heard from clinical experts on the Committee that approximately a third to a half of all centres are referring samples for testing to help manage the treatment of Crohn's disease, especially in people whose disease loses response to a TNF‑alpha inhibitor. The Committee noted that there is a lot of interest in using these tests to support decision‑making in Crohn's disease, and that clinicians find them useful. It was concerned however, that the complexities in interpreting the results without a defined cut‑off threshold (section 6.6) and the potential for misclassification (section 6.7) meant that tests could be incorrectly used by clinicians without specialist knowledge of the tests. The Committee therefore concluded that at this time, the number of laboratories using these tests should not increase beyond current numbers, unless the tests are used in the context of data collection or a research study.

The Committee considered the different scenarios for using the ELISA kits, that is, in people whose disease loses response to TNF‑alpha‑inhibitor treatment, and in people whose disease is responding to TNF‑alpha‑inhibitor treatment. The Committee noted the differences in the ICERs between the 2 scenarios, and that using the ELISA kits in people whose disease loses response to TNF‑alpha‑inhibitor treatment was associated with greater savings per QALY lost compared with using the ELISA kits in people whose disease is responding to TNF‑alpha‑inhibitor treatment (section 6.10). It also noted that there was less uncertainty in the cost savings in people whose disease loses response to TNF‑alpha‑inhibitor treatment compared with people whose disease was responding to TNF‑alpha‑inhibitor treatment (section 6.14). The Committee further noted that people with Crohn's disease with loss of response to TNF‑alpha inhibitors have a low quality of life (section 6.11) and limited treatment options (section 6.2). The Committee concluded that in people whose disease loses response to TNF‑alpha‑inhibitor treatment, the ELISA kits should be used in laboratories alongside data collection through a relevant registry or audit. The Committee also concluded that only laboratories that are currently using these tests and have expertise in immunoassay analysis and a thorough understanding of the technical factors that may affect the results should continue to use them. These laboratories should work closely in a network with the treating or referring clinician to ensure the appropriate use of the tests and interpretation of the results. The Committee also concluded that in people whose disease responds to TNF‑alpha‑inhibitor treatment, the ELISA kits should be used only in research.

The Committee considered the use of ELISA kits (LISA‑TRACKER, IDKmonitor, and Promonitor) in children with Crohn's disease and noted that no evidence on children was identified in the assessment. It heard from a clinical expert on the Committee that the effect of Crohn's disease on children can be slightly different to that in adults, for example, resulting in growth delay and psychiatric problems. It heard further that there is growing interest from paediatric clinicians in using these tests to help guide treatment in children with Crohn's disease. The Committee therefore encouraged data collection and further research into using ELISA kits (LISA‑TRACKER, IDKmonitor and Promonitor) to support decision‑making in children with Crohn's disease.

The Committee considered the advantages and disadvantages of concurrent testing and reflex testing. It heard from a clinical expert on the Committee that most centres in the UK use a concurrent‑testing strategy. The Committee noted that when a concurrent‑testing strategy is used, some tests may be wasted because samples with TNF‑alpha inhibitor present are unlikely to have free antibodies to TNF‑alpha inhibitor present (which is what most of the anti‑drug antibody ELISAs measure). However, a reflex‑test strategy may cause an unacceptable delay in giving results because fewer samples would be tested for antibodies to TNF‑alpha inhibitor and a laboratory would often wait for a full batch of samples before doing the test. The Committee noted that the ICER for the concurrent‑test strategies and the reflex‑test strategies were similar. It concluded that either test strategy could be used in research.

The Committee considered the research being conducted on tests to measure levels of TNF‑alpha inhibitors and antibodies to TNF‑alpha inhibitors in Crohn's disease. The Committee heard from a clinical expert that this is a fast‑moving area and a lot of research is being done. It also noted that the UK‑based PANTS – Personalised Anti‑TNF Therapy in Crohn's Disease – study should provide relevant data; but results are not expected until the end of 2016. The Committee concluded that data from this ongoing research are likely to be important when the guidance is considered for updating in the future.