Precision IFX: Using a Dashboard to Individualize Infliximab Dosage

Therapeutic monoclonal antibodies (MAbs) targeting the tumor necrosis alpha pathway (anti-TNFα, anti-TNF) in the treatment of immune diseases such as rheumatoid arthritis, psoriasis, and inflammatory bowel diseases (IBD) have improved short and long term clinical outcomes. Crohn's Disease (CD) and ulcerative colitis (UC), two main subtypes of IBD, are chronic diseases resulting from immune dysregulation in genetically susceptible individuals. CD and UC are conventionally treated using anti-inflammatory agents including aminosalicylate based therapies (mesalamines), corticosteroids, and antimetabolites such as purine analogs (azathioprine and 6-mercaptopurine) and methotrexate. A high percentage of patients fail to respond or are intolerant to these therapies and require treatment with anti-TNF. However, despite their therapeutic efficacy, approximately 20% of patients show no or limited response during induction therapy (primary non-responders) and in up to 50% of responders, treatment becomes ineffective during maintenance therapy despite initial response (secondary non-responders). Recent publications have underscored substantial variability in patient exposure and response when anti-TNF therapies are administered at the labeled induction and maintenance dose, supporting the need to individualize dosing to account for variability and ensure safe and sustainable efficacy. Suboptimal exposure can be attributed to under-dosing, rapid drug clearance and/or the development of anti-drug antibodies (ADA) and can result in primary or secondary loss of response (LOR). Identifying an individual's effective dose and adjusting the doses of anti-TNF over the course of treatment to maintain effective concentrations is not intuitive.

Software-guided dosing has been shown to effectively control doses for individual patients and to increase efficiency in clinics. Individualized adaptive dosing using PK models has been undertaken but was a labor-intensive process prior to using dashboard systems. Several dashboard systems already exist to improve dosing in pediatric patients. Clinical use of such systems is still limited, in part because of a lack of familiarity with dashboards, ineffective communication to practicing clinical staff on the use and benefits of such systems to facilitate decision making, and the resources required to use modeling to fully individualize treatment. However in the case of pediatric patients, particularly for those patients dosed based on body size (e.g. mg/kg or mg/m2), the drug exposure in pediatrics is often substantially lower than adult patients making these dosing metrics particularly difficult for patients with low body weight or pediatrics, as has been shown for infliximab. This suggests that pediatric patients would potentially garner the greatest benefit from individualized dosing.

Until recently, effective use of drug concentrations and biomarkers has been limited by the lack of decision support tools allowing physicians to integrate patient data and generate treatment recommendations. Implementation of adaptive Bayesian dosing in the clinic has not yet gained wide acceptance, and requires careful evaluation and testing. However Bayesian forecasting has been shown to substantially increase the number of patients whose trough phenytoin levels were within the target range and improve clinical outcomes in pediatric oncology patients. Van Lent-Evers at all found that the use of Bayesian adaptive dosing of aminoglycosides offered resulted in higher antibiotic efficacy, shorter hospitalization, and reduced incidence of nephrotoxicity. The authors also found lower treatment costs in patients who were dosed using Bayesian approaches. There have been a number of dashboard systems developed recently for improving treatment in pediatric oncology and infectious disease although these have not yet gained widespread clinical use.

From a clinician's perspective, it would be highly advantageous to be able to optimize the exposure of MAbs in patients receiving these therapies for disease control. The advent of commercially available drug and ADA concentration assays has indeed improved understanding of why patients are not responding or are losing response while in maintenance. With conventional weight based (on label) dosing, the investigators make the assumption that all patients clear anti-TNF therapies at the same rate and do not take into account inter-individual variability. A model that selects the correct induction dose based on clinical variables that alter clearance and allows clinicians to dose adjust in maintenance as the disease activity, weight, and drug concentrations fluctuate over the course of disease would not only be helpful but also economical. Consistent with the potential advantage of dashboards, a recent evaluation found individualized infliximab dosing reduced treatment costs compared to conventional dosing.

Presently for infliximab, patients are escalated from 5 mg/kg to 10 mg/kg and without necessarily considering that a patient may benefit from just increasing to 6 mg/kg to maintain therapeutic concentrations based on the dashboard predictions. Moreover it could be that a frequency change in dosing (e.g., from every 8 weeks to every 4 or 6 weeks) should be considered rather than dose escalation which carries significantly more cost than more frequent infusions. Rather than waiting for a patient to declare themselves a failure of the indicated dosing, predictive models can ensure the investigators are dosing correctly up front and being proactive and flexible with dosing regimens.

The ultimate goal of precision medicine is to utilize new information to optimize therapy for individual patients so that patients are treated with the right dose of the right drug at the right time. This approach is intended maximize benefit and minimize risk. Research has provided a wealth of new information, but health care providers are not always equipped to collect and manage this information in the patient care setting. Thus, dashboard systems may provide an important decision-support tool to facilitate the use of this new information into patient care.

The shift from conventional empirical dose adjustments to dashboard facilitated dosing will require access to the models developed during drug development or during post-marketing evaluations. However, prior to routine implementation in clinical use, dashboard systems will need to be designed to merge seamlessly with current clinical practice and the use of these systems will need to be verified by prospective clinical trials showing the benefit of this approach, and education about these systems will have to be made available to practicing physicians.

Preliminary data of a retroactive study completed using an IFX dashboard prototype showed a there was 0.70 concordance of actual trough values with forecasted trough values when using clinical profiles with laboratory observations from the first maintenance infusion. The dashboard retroactively evaluated dosing regimens for the patients and recommended every 7-8 week dosing in 56% of patients who received every 7-8 week. The dashboard system recommended a dose decrease for 52% of subjects and dose increased for 38% of patients. Additionally, 71.4% of subjects who developed antibodies were recommended dose changes and/or dosing frequency changes.

The dashboard system will actively monitor and dose to target a maintenance drug trough level after the standard induction period is complete. By proactively monitoring and dosing patients, this study hopes to reduce the frequency of subjects losing response due to anti-drug antibody development and to increase the frequency of subjects attaining and sustaining therapeutic drug. Overall, the Precision IFX dashboard will serve as a decision support tool for the clinician. The clinician will ultimately determine the appropriate dosing regimen to use for the patient. The dosing regimen will be within standard-of-care guidelines and within guideline and limits usually practiced in clinic and authorized by insurances.