Immunosuppressant adherence in adult outpatient hematopoietic cell transplant recipients

Abstract

Introduction:

Medication nonadherence continues to be challenging for allogeneic hematopoietic cell transplant (HCT) recipients. The risk and severity of chronic graft-versus-host disease (GVHD) are associated with low immunosuppressant concentrations (which can be improved with model-informed precision dosing (MIPD)) and with immunosuppressant nonadherence (which can be improved with acceptable interventions).

Methods:

With the goals of improving adherence and achieving therapeutic concentrations of immunosuppressants to eliminate GVHD, we characterized the feasibility of using the Medication Event Monitoring (MEMS^®) Cap in adult HCT recipients.

Results:

Of the 27 participants offered the MEMS^® Cap at the time of hospital discharge, 7 (25.9%) used it, which is below our a priori threshold of 70%. These data suggest the MEMS^® Cap is not feasible for HCT recipients. The MEMS^® Cap data were available for a median of 31 days per participant per medication (range: 7 to 109 days). The average daily adherence per participant ranged from 0 to 100%; five participants had an average daily adherence of over 80%.

Conclusions:

MIPD may be supported by MEMS^® technology to provide the precise time of immunosuppressant self-administration. The MEMs^® Cap was used by only a small percentage (25.9%) of HCT recipients in this pilot study. In accordance with larger studies using less accurate tools to evaluate adherence, immunosuppressant adherence varied from 0 to 100%. Future studies should establish the feasibility and clinical benefit of combining MIPD with newer technology, specifically the MEMs^® Button, which can inform the oncology pharmacist of the time of immunosuppressant self-administration.

Introduction

Allogeneic hematopoietic cell transplantation (HCT) has a delicate balance; the grafting of cells from one individual to another provokes immunologic reactions involved in the engraftment of the donor cells, graft-versus-host disease (GVHD), control of malignancy (termed graft versus tumor immunity, GVT), the development of tolerance, and immune reconstitution.¹ The primary goals of post-graft immunosuppression are preventing the development of GVHD while maintaining GVT and ensuring engraftment.¹ GVHD persists as a significant source of nonrelapse mortality.² There is substantial complexity with the outpatient GVHD prevention regimens (Table 1).

Table 1.

Examples of daily immunosuppressant pill count by GVHD prevention regimen for a typical adult (80 kg) HCT recipient. The number of pills at discharge assumes discharge on day 21.

	GVHD prevention		Recommended Dosing	Daily Pill Count
				Before first taper	Before second taper
Myeloablative conditioning	Regimen A – 8 pills at discharge
	CSP/MTX39	CSP	3 mg/kg/day IV or 5 mg/kg/dose PO divided Q12H, days –1 to +90, then taper 5% per week and discontinue by day +180	8 capsules (10 mg/kg/day × 80 kg = 800 mg rounded to 8 capsules of 100 mg)	NA
		MTX	15 mg/m2 IV, day +1, 10 mg/m2 IV, days +3, +6, +11	NA	NA
	Regimen B – 2 pills at discharge
	TAC/MTX39	TAC	0.03 mg/kg/day continuous IV infusion or 0.12 mg/kg/day PO BID, dose adjusted to TAC concentrations of 5–10 ng/mL, taper over 1 month beginning at day +30	2 capsules (0.12 mg/kg × 80kg = 9.6 mg rounded to 2 capsules of 5 mg)	NA
		MTX	see Regimen A	NA	NA
NMT – related donor	Regimen C – 12 pills (assumed all outpatient treatment)
	CSP/MMF39	CSP	see Regimen A	8 capsules (see Regimen A)	8 capsules (see Regimen A)
		MMF	15 mg/kg/dose IV or PO Q12H for HLA-matched related, taper over 1–3 months	4 capsules (15 mg/kg/dose × 2 doses/day × 80 kg= 1200 mg rounded to 2 capsules of 500 mg)
	Regimen D – 6 pills (assumed all outpatient treatment)
	TAC/ MMF39	TAC	See Regimen B	2 capsules (see Regimen B)	2 capsules (see Regimen B)
		MMF	See Regimen E	4 capsules (see Regimen C)
NMT – unrelated donor	Regimen E – 15 pills (assumed all outpatient treatment)
	CSP/MMF/SIR40	CSP	PO twice daily from day −3 and, in the absence of GVHD, tapered from day +96 through day +150	8 capsules (see Regimen A)	8 capsules (see Regimen A)
		MMF	15 mg/kg/dose IV or PO Q8H for unrelated donors, discontinued on day 40.	6 capsules (15 mg/kg/dose × 3 doses/day × 80 kg = 1200 mg rounded to 2 capsules of 500 mg)
		SIR	2 mg PO once daily from day −3 to +180	1 tablet (2 mg tablet)	1 tablet
RIC – Haploidentical	Regimen F – 8 pills at discharge
	BMT CTN 110141	CY	CY 50 mg/kg day +3 and +4
		TAC	adjusted to maintain a concentration of 5–15 ng/ml from day 5 through day 180.	2 capsules (see Regimen B)	2 capsules (see Regimen B)
		MMF	15 mg/kg/dose (max 1 gram) PO TID, day 5 to 35	6 capsules (See Regimen D)
	Regimen G – 14 pillsat discharge
RIC - Cords	BMT CTN 110141	CSP	Adjusted to maintain a trough level of 200–400 ng/ml through day 100 and tapered by 10% weekly until discontinued between day 180–200	8 capsules (see Regimen A)	8 capsules (see Regimen A)
		MMF	15 mg/kg PO TID (max dose of 3 gram/day) through day +35	6 capsules (see Regimen D)

Abbreviations: AIBW – adjusted ideal body weight; CSP – cyclosporine; CY – cyclophosphamide; GVHD – graft versus host disease; IV – intravenous, kg – kilogram; max – maximum; MMF – mycophenolate mofetil; NMT – nonmyeloablative transplant; PO – oral; Q8H – every 8 hours; Q12H – every 12 hours; RIC – reduced intensity conditioning; SIR – sirolimus; TAC – tacrolimus; TID – three times a day (assumed to be the same as every 8 hours); TBW – total or actual body weight

Low immunosuppressant trough (pre-dose) concentrations or exposure (i.e., area under the plasma concentration-time curve or AUC) is associated with a higher risk of GVHD.^{1, 3} Dosing some (i.e., cyclosporine, sirolimus, tacrolimus) immunosuppressants based on trough concentrations, often by the oncology pharmacist, has been the current standard of care for over 20 years. However, other immunosuppressants, such as mycophenolic acid (MPA), continue to be dosed based on body weight despite data suggesting that model-informed precision dosing (MIPD) can be beneficial. For example, among patients given nonmyeloablative (low dose) HCT with an unrelated donor graft, low MPA AUC predicted the severity of acute GVHD⁴, which is consistent with the findings of other investigators.^{5, 6} Low total MPA AUC also predicted high non-relapse mortality, potentially due to a higher risk of severe acute GVHD. To achieve higher MPA AUC, mycophenolate mofetil (MMF) doses should be increased for covariates associated with more rapid MPA clearance, specifically in patients receiving cyclosporine and in patients with lower albumin concentrations.⁷ Furthermore, population pharmacokinetic models can be used with limited sampling schedules for MIPD of MMF to a target MPA AUC. These approaches have been used with busulfan⁸ and cyclophosphamide⁹ in HCT recipients. In the renal transplant setting, MIPD dose personalization of MMF suggested lower rejection rates and gastrointestinal toxicity.^{10, 11} Two of the three studies used population pharmacokinetic-based limited sampling schedule (LSS) to personalize oral MMF doses, reflecting acceptance of these tools in the solid organ transplant community. However, such studies have yet to be completed in HCT over the past ten years after calling for studies in which oral MMF doses are personalized using a population pharmacokinetic-based LSS.⁴

Aberrant pharmacokinetics of insufficient dose personalization are not the only factors contributing to low immunosuppressant concentrations or AUC; nonadherence is an additional contributor.^{1, 3} Over the past thirty years, several publications have described nonadherence to outpatient medications in HCT recipients.^12–23 Specifically, 11.2%²³ to 64.6%¹⁹ of HCT recipients are nonadherent with immunosuppressants. In addition, nonadherence is associated with the development and severity of chronic GVHD.^{19, 22} Despite chronic GVHD being associated with nonadherence, there is a paucity of data regarding how to reduce nonadherence and improve personalized dosing of immunosuppressants.

Data regarding immunosuppressant dosing history is an important step in developing such strategies. To improve immunosuppressant dosing and reduce chronic GVHD, we evaluated the feasibility of measuring the date and time of each immunosuppressant intake using the Medication Event Monitoring System (MEMS^®, Aardex Group, Liège, Belgium) Caps in adult HCT recipients receiving outpatient immunosuppressants. The MEMS^® system is considered the most reliable method to monitor and record medication adherence.^{24, 25}

Methods

Study population.

From December 2018 to June 2019, participants were enrolled in this prospective study of using MEMS^® Caps to measure adherence to outpatient post-graft immunosuppression, which was a secondary aim to three prospective trials that were approved by the institutional Investigational Review Board. One of the three trials (Protocol 18387) evaluated the feasibility of a preconditioning HCT-modified geriatric assessment²⁶. The other two trials evaluated biomarkers for receiving post-graft immunosuppression with post-transplant cyclophosphamide (Protocol 17422 and 18358).^{27, 28} All participants were hospitalized; the attending physician determined their discharge date to the outpatient clinic. All participants provided written informed consent before study procedures; their HCT treatment was not affected by study participation. Demographic data were taken from the participants’ medical charts. Due to the limited sample size, only descriptive statistics are used.

Immunosuppressant adherence monitoring.

Medication adherence was assessed using MEMS^® Cap, an electronic pill cap that timestamps each cap opening that has been the ‘gold standard’ in adherence monitoring for over 30 years.²⁹ Participants were asked to use medication bottles with MEMS^® Caps to store and administer their medication for the duration of their immunosuppressive regimen. No interventions were made to influence immunosuppressant adherence. Feasibility was defined as the MEMS^® Cap use being completed in more than 70% of participants.²⁶

At the time of the study’s conduct, there was no structured adherence support for post-graft immunosuppression in the outpatient clinic.²⁵ The typical standard of care assessments after discharge to the outpatient clinic were, in general, that outpatients have follow-up clinic appointments twice a week and then with decreasing frequency to HCT day +100. Inpatients and outpatients are cared for by an interdisciplinary team consisting of physicians, mid-level providers (physician assistant/associate and advanced registered nurse practitioner), nurses, and psycho-oncologists. Other services, such as clinical pharmacy, physiotherapy, and social services, are included mainly on an inpatient basis. Before discharge, most inpatients receive an educational session with the inpatient pharmacist on preparing the medication at home without discussing specific strategies to enhance their medication adherence. During follow-up, nurses regularly assess adherence to immunosuppressants.²⁵

Data analysis of MEMS^®.

A priori decision rules were applied for periods when medication(s) were not dispensed from the bottle. Such periods were defined as “nonmonitored” and included: inpatient hospitalization, visits to the emergency room, or medication hold. In addition, data were coded as “nonmonitored” if families reported no longer using the MEMS^® Cap; the participant or family members did not provide reasons for discontinuation.

The data were analyzed similarly to that of McGrady and colleagues¹⁷, which was the only publication of MEMS Cap^® of medication adherence in HCT recipients. This allowed us to evaluate if adherence has changed in HCT recipients since 2014.¹⁷ A detailed description of the data analysis process is described in Methods S1. For each participant, the average daily adherence was calculated by dividing the number of pill cap openings per day by the number of prescribed doses for that day. The frequency and average length of medication interruptions (>24 h between doses) were calculated when daily medications were prescribed.

Clinical outcomes.

The assessment of acute GVHD and chronic GVHD was described previously.³⁰

Results

Participant characteristics.

Table 2 contains a summary of the participants’ pre-HCT characteristics. The median age was 48.8 years (range: 19.4 – 72.6), and all underwent HCT for a hematologic malignancy. The majority (n=4) received an HLA-haploidentical donor graft.

Table 2.

Participant characteristics (n=7)^a

Age (years)	48.8 (19.4–72.6)
Male Sex	7 (100%)
Race – Caucasian	7 (100%)
Ethnicity
Non-Hispanic	4 (57.1%)
Hispanic	3 (43.9%)
HLA Category/Donor Typeb
Haploidentical	4 (57.1%)
Unrelatedb	2 (28.6%)
Relatedb	1 (14.3%)
Diagnosis
Acute Myeloid Leukemia	3 (43.9%)
Acute Lymphoblastic Leukemia	2 (28.6%)
Hodgkin’s Lymphoma	1 (14.3%)
Myeloproliferative Neoplasm	1 (14.3%)
HCT Conditioning Regimen
Fludarabine/melphalan ± radiation	3 (42.8%)
Fludarabine/cyclophosphamide ± radiation	2 (28.6%)
Radiation ± fludarabine	2 (28.6%)
GVHD Prophylaxis
Tacrolimus/PTCY ± mycophenolate mofetil	5 (71.4%)
Tacrolimus/sirolimus ± ruxolitinib	2 (28.6%)

^aData presented as median (range) or n (%)

^bone mismatched and one matched unrelated donor; one matched related donor

Immunosuppressant adherence.

The STROBE diagram is shown in Figure 1. Of the 27 participants offered the MEMS^® Caps, only 7 (25.9%) used it. This is less than our a priori feasibility definition, which was MEMS^® Cap use being completed to the end of immunosuppressant treatment in greater than 70% of participants.²⁶ Approximately half of the participants who were offered the MEMS^® Cap agreed to use them; twelve participants refused the MEMS^® Cap because they were unwilling to change their established medication routine. The MEMS^® Cap was brought back but then lost in the outpatient clinic for at least one participant.

Figure 1.

Figure 1: STROBE Diagram of MEMS^® Cap

Of the seven participants who used the MEMS^® Cap, the MEMS^® Cap medications used were: tacrolimus (TAC, n=7), mycophenolate mofetil (MMF, n=3), and sirolimus (SIR, n=1). Seven participants completed their MEMS^® monitoring for the planned duration of their immunosuppressant medication prescription. The MEMS^® Cap data were available for a median of 31 days per participant per medication (range: 7 to 109 days). The average daily adherence ranged from 0 to 100%; five participants had an average daily adherence of over 80% (Table 3). There were numerous days when an immunosuppressant dose was missed and numerous interruptions at varying times and durations. The occurrence of GVHD is also reported for descriptive purposes.

Table 3.

Immunosuppressant adherence assessed with MEMS^® Caps.

Participant ID	1	2	3		4		5		6		7
Medication monitored	TAC	TAC	TAC	MMF	TAC	MMF	TAC	MMF	SIR	TAC	TAC
Administration frequency at discharge	BID	BID	BID	BID	BID	BID	BID	BID	QD	BID	BID
Expected medication administration (days)	22	109	102	66	96	23	7	14	48	35	11
# of days with correct medication administration	18	97	0	49	17	1	6	14	7	0	1
Number of changes in medication	2	2	1	1	4	0	0	0	5	0	0
Per participant, average daily adherence for whole time period of medication	81.8%	89.0%	0%	74.2%	17.7%	4.35%	85.7%	100%	14.6%	0%	9.09%
# of days with missed doses	3	1	102	14	78	23	0	0	40	35	10
# of days where the additional doses were taken	1	11	0	0	1	0	1	0	1	0	0
# of medication interruptions	2	1	2	7	4	2	0	0	7	1	2
Length of interruptions in days	1.5	0.5	100.5	7	76.5	19	0	0	41	28	1
Did this participant have GVHD?	No	Yes	Yes		Yes		Yes		Yes		Yes
acute GVHD (worse grade)		II	I		I		I		No		I
chronic GVHD		No	No		No		No		Mild		Severe

Abbreviations ; QD – once a day; BID – two times a day. MMF – mycophenolate mofetil; SIR – sirolimus; TAC – tacrolimus

Discussion

MIPD improves the effectiveness and safety of HCT conditioning regimens.^{8, 9} In general, low immunosuppressant concentrations are associated with clinical outcomes, leading us to hypothesize that GVHD rates may be lowered by dose personalization using MIPD^{4, 31} in combination with the MEMs^® Cap. Thus, we took a first step towards characterizing adherence with oral outpatient immunosuppressants. Characterizing adherence can inform 1) population pharmacokinetic-guided LSS for evaluating immunosuppressant pharmacokinetics and MIPD;^{31, 32} and 2) medication nonadherence interventions.^{1, 3, 31} We hypothesize that GVHD rates may be lowered by identifying when immunosuppressant nonadherence occurs, allowing rapid intervention behavioral interventions with graphene-based biosensors and MIPD to achieve the desired target immunosuppressant concentration or exposure (Figure 2).

Figure 2.

Proposed study design to lower GVHD rates. HCT recipients should receive a multi-pronged intervention with point-of-care monitoring of adherence (MEMS^® Button) and with point-of-care monitoring of immunosuppressant concentrations (novel wearables) with MIPD to guide when to use an integrated facilitated care model³⁸ to lower GVHD.

In allogeneic HCT recipients, nonadherence rates for oral medications range between 21%−66%,²³ with the higher percentages found in longer-term survivors.^{14, 18, 19, 21, 22, 33} A 2017 meta-analysis found limited research in HCT medication adherence, identifying five studies with highly variable methods. The meta-analysis had a conclusion that nonadherence commonly occurs (5.3% to 67%) and is an important area for research and intervention.³³ In our pilot study, average daily adherence per participant for whole time period ranged from 0 to 100%; we could not evaluated monthly adherence because of the short duration of MEMs^® Cap use by our participants (Table 3). The only other MEMs study in HCT recipients reported that six adolescents (ages 12–18 years) had a daily adherence ranging from 58 to 92%. These data suggest that nonadherence persists in HCT recipients, although both sample sizes are limited. We calculated adherence using the same methods of McGrady et al (2013) to allow for this comparison of medication adherence in HCT recipients over the past ten years. However, considering the heterogeneity and low accuracy of medication adherence measures, future studies should use the TEOS (Timelines-Events-Objectives-Sources) framework which was recently developed as a structured approach to formulate operational definitions of adherence.^{34, 35}

Prospective, longitudinal studies on medication adherence were recommended to characterize further the effect of nonadherence on clinical outcomes, such as GVHD and mortality.³³ Prospective and longitudinal studies can be facilitated by electronic monitoring devices, which are suggested to be the gold standard for measuring adherence in research,³³ that measures the date and time of each medication intake are needed. Traditionally, such an electronic monitoring system must be brought to the hospital for care personnel to read the data,²⁹ which led to logistical obstacles that hindered the assessment of immunosuppressant adherence (Figure 1). Recently, Ribault and colleagues reported that focus groups of six adult HCT recipients (6 weeks to 2 years after allogeneic HCT) suggested that the MEMS^® Button was most acceptable and preferred compared to the MEMS^® Helping Hand and the MEMS^® Cap.²⁵ After completion of this study, the MEMS^® Button with Bluetooth^® Capability became available, which may overcome the majority of obstacles that we identified (Figure 1).²⁵ Based on our experience, the MEMS^® Button with Bluetooth^® Capability has two advantages. First, it has a clip that allows the MEMS^® Button and the immunosuppressants to be kept together; 12 of the 27 patients refused to use the MEMs^® Cap because they had an established medication routine (often involving pill boxes) that they were unwilling to change (Figure 1). Second, the MEMS^® Button has Bluetooth^® Capability, which allows for collecting the adherence data without returning it to the clinic; 1 of the 14 participants had their MEMs^® Cap lost in the clinic (Figure 1), but our clinical research staff also had to spend an appreciable amount of time finding the MEMs^® Cap from other participants in which we obtained data. Nearly a half of the participants refused the MEMS^® Cap because they were unwilling to change their established medication routine. This barrier may be overcome with the MEMS^® Button, which can be used the participant’s existing routine for medication self-administration. Of those 14 participants who used the MEMS^® Cap, one was lost in the outpatient clinic. This situation should not occur with the Bluetooth^® Capability of the MEMS^® Button.

Based on our experience in this pilot study, we propose including MIPD with the MEMS^® Button to personalize the immunosuppressant doses in those patients with low immunosuppressant trough concentrations or AUC (Figure 2).^{4, 31, 32} Furthermore, these technologies can be paired with various strategies to improve medication adherence, which range from technology-guided adherence interventions, education, simplification of medication regimens, reminders, cognitive behavioral interventions, social support, follow-up by health care providers, and insurance incentives.^{33, 36} For example, one method to simplify the GVHD prevention medication regimen is to decrease the number of pills administered per day. For example, as shown for an 80 kg patient in Table 1, Regimen F (post-transplant cyclophosphamide, tacrolimus, and MMF) has 8 pills per day while Regimen E (tacrolimus, MMF, and sirolimus) has 15 pills per day. We hypothesize that GVHD rates can be lowered by improving adherence and by personalizing the immunosuppressant dose through point-of-care technologies and MIPD (Figure 2). Furthermore, the recent development of novel, fully integrated point-of-care wearable or portable graphene-based biosensors³⁷ may facilitate point-of-care pharmacokinetic sampling for immunosuppressants (Figure 2). We propose GVHD rates can be lowered by the multifactorial intervention with two foci. They are, first, improving adherence through the MEMS^® Button and electronic health-enabled integrated facilitated care model³⁸; and, second, using novel point-of-care devices for quantitation of immunosuppressant concentrations,³⁷ LSS feasible in the outpatient setting,³¹ and MIPD³². We have successfully developed such LSS for MPA after oral MMF administration that was acceptable for study participation in a pharmacodynamic study in nonmyeloablative HCT recipients.³¹

Conclusion

MIPD may be supported by MEMS^® technology to provide the precise time of immunosuppressant self-administration. However, only a small fraction (7 of 27) of HCT recipients used the MEMs^® Cap. Immunosuppressant adherence varied from 0 to 100% in our pilot study, which agrees with larger studies using less accurate tools to evaluate adherence. Future studies should establish the feasibility and clinical benefit of combining the MEMs^® Button with MIPD.