Adherence to oral hormonal therapy in advanced prostate cancer: a scoping review

Abstract

Background:

Orally administrated agents play a key role in the management of prostate cancer, providing a convenient and cost-effective treatment option for patients. However, they are also associated with adherence issues which can compromise therapeutic outcomes. This scoping review identifies and summarizes data on adherence to oral hormonal therapy in advanced prostate cancer and discusses associated factors and strategies for improving adherence.

Methods:

PubMed (inception to 27 January 2022) and conference databases (2020–2021) were searched to identify English language reports of real-world and clinical trial data on adherence to oral hormonal therapy in prostate cancer using the key search terms ‘prostate cancer’ AND ‘adherence’ AND ‘oral therapy’ OR respective aliases.

Results:

Most adherence outcome data were based on the use of androgen receptor pathway inhibitors in metastatic castration-resistant prostate cancer (mCRPC). Self-reported and observer-reported adherence data were used. The most common observer-reported measure, medication possession ratio, showed that the vast majority of patients were in possession of their medication, although proportion of days covered and persistence rates were considerably lower, raising the question whether patients were consistently receiving their treatment. Study follow-up for adherence was generally around 6 months up to 1 year. Studies also indicate that persistence may drop further with longer follow-up, especially in the non-mCRPC setting, which may be a concern when years of therapy are required.

Conclusions:

Oral hormonal therapy plays an important role in the treatment of advanced prostate cancer. Data on adherence to oral hormonal therapies in prostate cancer were generally of low quality, with high heterogeneity and inconsistent reporting across studies. Short study follow-up for adherence and focus on medication possession rates may further limit relevance of available data, especially in settings that require long-term treatment. Additional research is required to comprehensively assess adherence.

Introduction

Prostate cancer is a hormone responsive cancer with over 1.4 million new cases resulting in nearly 400,000 deaths worldwide in 2020.¹ Initial systemic therapies for advanced prostate cancer consisted of subcutaneously/intramuscularly administered androgen deprivation therapy (ADT), which decreases testosterone production to castrate levels by modulating the production and/or activity of the luteinizing hormone-releasing hormone (LHRH) using LHRH agonists such as goserelin, leuprolide, and triptorelin or LHRH antagonists such as degarelix.^2–4 These have been standard therapy for advanced prostate cancer since the 1980s. When administered in early settings, treatment can be ongoing for prolonged periods^5,6 and as such, ADT is usually administered as long-acting depot formulations which can range from a 1-month interval to up to a year.^7–13 More recently, the first orally administered LHRH antagonist relugolix¹⁴ was approved by the FDA on 18 December 2020.¹⁵

Androgen receptor-axis-targeted therapies (ARATs) were the first major class of oral hormonal agents to be approved for advanced prostate cancer. Bicalutamide, the first-generation nonsteroidal androgen receptor inhibitor, was approved by the FDA for use alone or in combination with ADT as early as 1995,^16,17 with the oral 17α-hydroxylase/C17,20-lyase (CYP17)-targeted androgen synthesis inhibitor abiraterone currently administered orally at a dose of 1000 mg (two 500 mg tablets or four 250 mg tablets) once daily with prednisone 5 mg twice daily was first approved in 2011¹⁸ and the second-generation androgen receptor inhibitor, enzalutamide currently administered at a dose of 160 mg (two 80 mg or four 40 mg tablets) once daily,¹⁹ both originally for metastatic castration-resistant prostate cancer (mCRPC) previously treated with docetaxel-containing chemotherapy. Median duration of therapy was 8 months for abiraterone²⁰ and 8.3 months for enzalutamide²¹ in the pivotal trials. Starting in 2018, enzalutamide was approved for non-mCRPC^22,23 followed by approvals of next-generation androgen receptor inhibitors apalutamide^22,24 and darolutamide²⁵ and in 2019 enzalutamide²⁶ and apalutamide²² were approved for high-risk metastatic castration-sensitive prostate cancer.^2,4,27

The use of oral agents is often preferred by many cancer patients^28–30; however, use is often associated with issues of adherence that can compromise therapeutic outcomes. Adherence is generally measured by either subjective measures based on self-reported assessments through questionnaires or diaries or objective means based on medication counts from observer-reported assessments (Table 1). While administration of subcutaneous and intravenous agents is closely monitored by highly trained medical professionals and accompanied by extensive education, a transition to oral therapies often means a shift in responsibility for the acquisition, administration, education, and safety monitoring of medications from healthcare professionals to patients and their support networks.^31–34 This additional responsibility may present a challenge for the average cancer patient, particularly as it relates to agents that are less well tolerated, have complex schedules, or require frequent administration.^33,35–39 Multiple demographic (e.g. age, ethnicity, and gender), socioeconomic (e.g. education and income levels), and other patient-related factors (e.g. comorbidities and forgetfulness) as well as disease- (e.g. stage and symptoms) and treatment-related (e.g. duration, toxicity, dose, and frequency) factors have been assessed in studies of adherence to oral anticancer therapies with only a few showing consistent associations across multiple studies in different cancers.^{36–38,40,41}

Table 1.

Treatment adherence and persistence measures for oral hormonal therapies in advanced prostate cancer.

Measure	Brief description	Advantages and disadvantages
Self-reported assessments42,43		Easy and inexpensive to administer; capture medication adherence behavior which can provide actionable information; assesses drug omission errors; high specificity
		Limited by patient recall and tends to overestimate adherence due to patient reporting bias; low sensitivity
MMAS-444–47	Questionnaire instrument composed of four items with dichotomous yes/no response categories	Validated and widely used in a variety of patient settings; designed to avoid acquiescence bias commonly seen with chronic care patients; higher degree of concordance with objective measures; short with low response burden
		Limited assessment of nonadherence factors; suboptimal psychometric (quantification) properties
BAAS48–50	Structured questionnaire instrument initially developed for immunosuppressive drugs in transplant recipients and composed by four dichotomous items and a visual analog scale to measure overall adherence	Easily adapted to assess medication adherence in other chronically ill patient populations; assesses different aspects of adherence (dosing, timing, and frequency)
		Not validated in oncology settings
ACI51	Custom questionnaire instrument composed by 10 items with some derived from previously validated instruments of adherence (ASK-12, BMQ-Specific, BAAS, and MMAS)	Allows for more comprehensive assessment of adherence-related factors (such as cost and AEs), according to individual study goals
		Not validated
Observer-reported assessments42		Measures are generally simple and inexpensive to assess; not affected by patient reporting bias
		Does not directly capture information related to reasons for nonadherence; unable to directly assess medication ingestion
MPR42,44,52	Ratio of the summation of the ‘days’ supply’ of medication refills over a fixed or flexible interval; determined using prescription drug plan and pharmacy claim data	Simple, highly accessible, and widely used
		Does not consider early refills, refill gaps, or treatment discontinuation and may overestimate the true rate of medication adherence; variation in definitions used across different studies
PDC42,52–54	Calculated based on the fill dates and days’ supply for each fill of a prescription over a fixed measurement period; it differs from the MPR in that is not a simple summation of the days’ supply based on number of refills but rather a summation of the days covered in each refill cycle; determined using prescription drug plan and pharmacy claim data	Newer and extensively used method; results in more conservative estimates that consider discontinuation and timing of refills; operationally defined more consistently
		Measurement is more involved, requiring tracking of individual refill dates and determination of coverage in each refill cycle
Persistence rate42,53	Percentage of patients with no major treatment gaps between adjacent refills at a fixed timepoint	It is a better measure of how well patients can continue treatment as it is based on major treatment breaks and quantified using a patient ratio
		Does not consider small treatment breaks which may be relevant in treatment plans with the goal of continued biochemical control; does not discern between compliance-related and unrelated discontinuation events (e.g. due to disease progression)
Pill or dose count rates – AR/TA14,44,48	Percentage of expected compared to actual doses taken at different timepoints; commonly determined using pill count but can also be estimated from self-reporting	Simple, inexpensive, and objective
		It requires direct or indirect determination of number of pills or doses taken; pill or dose count is unfeasible in drugs with non-discrete dosages; may result in adherence underestimation

ACI, Adherence Composite Index; AEs, adverse events; AR, adherence rate; ASK-12, Adherence Starts with Knowledge 12; BAAS, Basel Assessment of Adherence Scale; BMQ, Belief about Medicines Questionnaire; MMAS, Morisky Medication Adherence Scale; MMAS-4, 4-item Morisky Medication Adherence Scale; MPR, medication possession ratio; PDC, proportion of days covered; TA, treatment adherence.

Hormonal therapy has formed the backbone of treatment for mCRPC for decades, and as oral hormonal therapy moves into the castrate-sensitive and nonmetastatic settings, it is unclear how this will affect adherence rates. These settings are associated with longer treatment durations where continued adherence is particularly important to achieve and maintain biochemical control. Although maintenance of testosterone levels below the castrate threshold (ideally set at 20 ng/dL) is a therapeutic goal across the disease spectrum, it has been particularly well established in the castrate-sensitive setting where testosterone escapes above the castrate level have been associated with poor outcomes.^55–60 Given the importance of maintaining high levels of adherence in prostate cancer, this review will summarize clinical trial and real-world adherence data for oral hormonal agents used in prostate cancer and critically appraise adherence measures, associated factors, and strategies to improve adherence to oral hormonal therapy in these settings.

Methods

A search of published and presented literature was conducted to identify real-world and clinical trial data on adherence to oral hormonal therapy in prostate cancer. PubMed (inception to 27 January 2022), the proceedings from the 2020 and 2021 American Society of Clinical Oncology (ASCO), and the European Society for Medical Oncology (ESMO) annual meetings as well as the ASCO Genitourinary Cancers Symposium were searched using the key search terms ‘prostate cancer’ AND ‘adherence’ AND ‘oral therapy’ OR ‘respective aliases’. A supplemental bibliographic search of review articles and pooled/meta-analyses was also conducted.

English language records were vetted at abstract level and confirmed at full text as needed by an initial reviewer (AP) and confirmed by a second independent reviewer (IM). Excluded records were non-original research articles, surveys, modeling and simulation studies, preclinical, correlative science, epidemiologic or biomarker studies, or case series or single case reports. Studies not specific to prostate cancer, not assessing oral hormonal agents, or without adherence outcomes were also excluded. Data were extracted to tables from full-text sources of eligible studies by a primary reviewer and accuracy confirmed by a second independent reviewer. Mean values and ⩾80% patient rates at the longest follow-up time available were preferentially extracted. Adherence values reported as ratios were converted to percentages and nonadherence rates to adherence rates.

Findings

The literature search identified a total of 694 records, resulting in a total of 16 eligible studies reporting results from assessment of adherence to oral therapies in prostate cancer patients (Preferred Reporting Items for Systematic Reviews and Meta-Analyses, Figure 1). In all, 13 observational studies assessed adherence and related measures exclusively in mCRPC patients (Table 2),^{48,51,61–71} while one phase III trial and two observational trials assessed adherence in advanced prostate cancer (Table 3).^14,53,72 Studies were mostly conducted in North America (n = 7) and Europe (n = 8) based on data from large national or provincial registries or hospital network databases (11), or smaller multi- or single-center analyses (5). Most prospective studies used questionnaire and survey tools to assess adherence (n = 4), while objective measurement of adherence rates based on pill counting was performed in two prospective trials. Prospective, randomized studies assessed the impact of adherence-enhancement measures (AEMs), administration (different doses and fasting versus low-fat meal), and different agents on adherence. Retrospective studies invariably used observer-reported measures based on analysis of prescription order or claim data (n = 11). Abiraterone and enzalutamide were by far the most common oral agents assessed for adherence. Among the high number of observational studies, reporting was nonstandard and data quality was generally low, limiting the strength of any conclusions.

Figure 1.

PRISMA diagram of eligible studies.

^aPrimary or associated reports of eligible studies that were not identified through database search. ASCO, American Society of Clinical Oncology; ASCO-GU, American Society of Clinical Oncology Genitourinary Cancers Symposium; CRPC, castration-resistant prostate cancer; ESMO, European Society for Medical Oncology; mCRPC, metastatic castration-resistant prostate cancer; n, number of reports or studies; PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Table 2.

Clinical studies assessing adherence to oral hormonal therapies in mCRPC.

First author; study type; time range; institutions/databases	Treatment setting; patient eligibility	Treatment regimen(s)	n	Median age (years)	Median follow-up (mos) [range]	Measures to improve compliance/adherence	Medication measure	Adherence rate	Disc. rate	Other adherence-related measures
Self-reported measures
Heiss et al.51 Prospective, multicenter, open-label, randomized trial 2012–NR Six centers in the United States and Singapore	mCRPC Diagnosed with mCRPC without prior use of abiraterone, enzalutamide, or other potent androgen pathway-targeted therapy	1000 mg abiraterone fasting	11	69.6	NR	NR	Self-reported in survey	Mean ACI Baseline (60 d): 98.2% Follow-up: 91.7% Baseline versus follow-up: p = 0.008	NR	NR
		250 mg abiraterone with a low-fat meal	19	73.2	NR	NR	Self-reported in survey	Mean ACI Baseline (60 d): 96.5% Follow-up: 97.9% Baseline versus follow-up: p = 0.35	NR	NR
Suttmann et al.71 Multicenter prospective 2-arm CRT October 2013–June 2018 87 German urology or oncology ambulatory care facilities	mCRPC Diagnosed with mCRPC who were either progressive after docetaxel-based chemotherapy or asymptomatic/mildly symptomatic after failure of ADT	Abiraterone plus prednisone with AEM	360	Mean: 75.3	NR	Educational video Calls by a study nurse Patient diary Dosage card Reminder SMS service	Self-reported	MMAS-4 score high 3 mos: 93.6% 6 mos: 92.9%	3 mos: 9.0% 6 mos: 14.3%	Median duration of treatment: 10 mos
		Abiraterone plus prednisone without AEM	315	Mean: 74.1	NR	No adherence measures	Self-reported	MMAS-4 score high 3 mos: 90.9% 6 mos: 92.6%	3 mos: 7.3% 6 mos: 11.2%
De la Llave et al.66 Multicenter survey study, ADOPTA. NR Eight centers across Spain	mCRPC Patients receiving either abiraterone or enzalutamide	Abiraterone or enzalutamide	50	Mean: 72	NR	NR	Self-reported in survey	Adherence = 100%: 84% Adherence > 75%: 100%	NR	Mean duration of treatment: 9.5 mos
Observer-reported measures
Banna et al.48 Prospective single-institution cohort April 2014–December 2018 Cannizzaro Hospital, Catania, Italy	mCRPC Diagnosed with PCa with increased PSA and castrate level of serum testosterone <50 ng/dL following ADT and initiation of treatment with abiraterone or enzalutamide	Abiraterone	22	76	26.8	NR	Pill counting	ARa 96.8%, p = 0.09	NR	ARa based on BAAS: 99.3%
		Enzalutamide	36	76	26.3	NR	Pill counting	ARa 93.6%	NR	ARa based on BAAS: 98.2%
Doshi et al.67 Retrospective national database analysis January 2012–December 2013 Medicare Chronic Condition Data Warehouse (CCW) Parts A, B, and D, United States	mCRPCb Patients with diagnosis of PCa and a claim of oral anticancer drug in the index period	Abiraterone	5585	NR	12	NR	Prescription orders	Mean interval-based (‘traditional’) PDC: 69% PDC ⩾ 80%: 47%	39%	NR
		Enzalutamide	1413		NR	NR	Prescription orders	Mean interval-based (‘traditional’) PDC: 71% PDC ⩾ 80%: 51%	33%	NR
Fallara et al.68 Retrospective national registry analysis January 2015–October 2019 The National Prostate Cancer Register, The Prescribed Drug Registry, and The Patient Registry, Sweden	mCRPCc Diagnosed with PCa until 31 December 2018 and who had filled a first prescription for abiraterone or enzalutamide in the study period	Abiraterone	1803	76.7	NR	NR	Prescription orders	MPR: 97% MPR ⩾ 80%: 95%	NR	Median time on treatmentd: 4.9 mos (IQR 2.6–11.7)
		Enzalutamide	4534	76.1	NR	NR	Prescription orders	MPR: 95% MPR ⩾ 80%: 92.0%	NR	Median time on treatmentd: 8.0 mos (IQR 3.6–16.4)
Caram et al.64 Retrospective large, multi-hospital registry analysis July 2013–June 2015 Medicare part D enrollees, United States	mCRPC Diagnosed with advanced PCa who survived and had sustained eligibility for at least 6 mos following their first fill	Abiraterone/enzalutamide	4153	NR	6	NR	Prescription orders	PDC ⩾ 80% over 6 mos: 75% 69% versus 76% (>85 versus <70 years of age, p < 0.01)	NR	NR
Lafeuille et al.69 Retrospective large, multi-hospital registry analysis December 2010–August 2012 Database 1: Truven Health Analytics Market Scan, United States June 2009–March 2013 Database 2: Symphony Health Solutions’ ProMetis Lx, United States	mCRPC Dataset 1: At least one PCa diagnosis and one metastatic diagnosis as well as at least one prostate cancer diagnosis after the index date	Abiraterone	Dataset 1: 515	Mean: 72.2	6 to ⩾12	NR	Prescription fill	Mean MPR: 93–96%e	NR	NR
	Dataset 2: Patients with at least two quarters of clinical activity prior to the index period and continuous quarters of clinical activity post-index with at least one PCa diagnosis	Abiraterone	Dataset 2: 3228	Mean: 72.2		NR	Prescription fill	Mean MPR: 93%f	NR	NR
Behl et al.63 Retrospective large, multi-hospital registry analysis October 2012– December 2014 Truven Health MarketScan database, United States	mCRPC Diagnosed with mCRPC who initiated treatment with abiraterone acetate or enzalutamide	Abiraterone	2591	NR	9.6	NR	Prescription orders	@12 mos Mean MPR: 95%, p = 0.0330 MPR ⩾ 80%: 91.7%, p = 0.0039	NR	Persistence (12 mos)g: 50.0%, p = 0.24 Median exposure to treatment: 5.7
		Enzalutamide	807	NR	6.6	NR	Prescription orders	@12 mos Mean MPR: 92 MPR ⩾ 80%: 85.4%	NR	Persistence (12 mos)g: 45.7% Median exposure to treatment: 4.0
Al-Ali et al.61 Retrospective large, multi-hospital registry analysis September 2013–August 2016 Database of the largest public insurance company in Austria	mCRPC Diagnosed with CRPC with at least one prescription for abiraterone and/or enzalutamide	Abiraterone	195	Mean: 74.4	NR	NR	Prescription orders	Mean MPR: 90%	NR	Mean duration of treatment: 9.5 mos
		Enzalutamide	139		NR	NR	Prescription orders	Mean MPR: 85%	NR	Mean duration of treatment: 7.6 mos
		Abiraterone/enzalutamide sequence	123		NR	NR	Prescription orders	Mean MPR: 88%	NR	Mean duration of treatment: 18.1 mos
Al-Ali et al.62 Retrospective large, multi-hospital registry analysis November 2011–December 2014 Wiener Gebietskrankenkasse database, Austria	mCRPC Diagnosed with CRPC with at least one prescription of abiraterone	Abiraterone	270	74	NR [1–38]	NR	Prescription fill	Mean MPR: 94.8% MPR ⩾ 80%: 89.3%	NR	Mean duration of treatment: 9.8 mos
Cindolo et al.65 Retrospective large, multicenter registry analysis January 2011–July 2017 ApotekeGold – administrative database of local health unit (LHU 02 Abruzzo), Italy	mCRPC Patients who started on LHRH analogs, LHRH antagonists, enzalutamide, and abiraterone	Abiraterone	109	77	NR	NR	Prescription orders	Median MPR: 99.0h	NR	Median persistencei: 7.8 mos
		Enzalutamide	14		NR	NR	Prescription orders	Median MPR: 95.5h	NR	Median persistencei: 8.5 mos
		LHRH analogs	1075		NR	NR	Prescription orders	Median MPR: 93.0h	NR	Median persistencei: 24.5 mos
		LHRH antagonists	80		NR	NR	Prescription orders	Median MPR: 97.0h	NR	Median persistencei: 14.9 mos
Smith et al.70 Retrospective provincial registry analysis August 2011–October 2013 Saskatchewan Cancer Agency, Canada	mCRPC Patients that received at least one dispensation of abiraterone from a SCA pharmacy	Abiraterone	86	NR	6	NR	Prescription orders	Mean MPR: 89.6% ± 21.2% Median MPR: 100% MPR ⩾80%: 82.6% MPR ⩾90%: 79.1%	Early non-persistence occurred infrequently: it was responsible for only 14.0% of all nonadherence at 6 mos and 18.4% at 1 year	Of patients with available follow-up to 1 year, 81.6% (31 of 38) maintained optimal adherence during the entire period

Studies were included in the table if they reported adherence rate or equivalent measures. Ordered by type of main measure, level of evidence (prospective, retrospective cohort studies, and surveys), and then trial size.

^aReported at source as nonadherence rates for the whole study period based on prescribed/taken pills ratio. Adherence rates were calculated by subtracting the reported non adherence rates from 100%.

^bAlthough study eligibility only explicitly required ‘metastatic’ prostate cancer patients, it was assumed that the vast majority of these prescription claims were for the approved mCRPC indication during the time covered by the study.

^cThe only indication for abiraterone and enzalutamide was mCRPC until 15 June 2018 when abiraterone was licensed for men with metastatic hormone-sensitive PCa. Thus, the authors assumed that men who had filled a prescription for abiraterone or enzalutamide were in mCRPC state, with the exception of men who had filled a prescription for abiraterone less than 6 months after PCa diagnosis, who were excluded.

^dUsing 30 days as the maximum time interval between two consecutive fillings without drug supply.

^eRange of MPRs reported across observation periods of different lengths: 6 months, at least 9 months, and at least 12 months.

^fVariable observation period.

^gDefined as a gap of more than 60 days between two adjacent refills or between the end of the last refill and the end of the evaluation period.

^hCalculated by days covered by therapy (number of defined daily doses delivered) by total persistence.

ⁱCalculated as the difference in days of the date of first and last delivery of drug or until the end of available data.

ACI, Adherence Composite Index; ADT, androgen deprivation therapy; AEM, adherence-enhancement measures; AEs, adverse events; AR, adherence rate; BAAS, Basel Assessment of Adherence Scale; CRT, cluster-randomized trial; d, days; Disc., discontinuation; IQR, interquartile range; LHRH, luteinizing hormone-releasing hormone; mCRPC, metastatic castration-resistant prostate cancer; MMAS-4, Morisky Medication-Taking Adherence Scale-4 item score; mos, months; mPCa, metastatic prostate cancer; MPR, medication possession ratio; n, number of patients; NA, not applicable; NR, not reported; PCa, prostate cancer; PCBaSE, Prostate Cancer Database Sweden; PDC, proportion of days covered; PSA, prostate-specific antigen; SCA, Saskatchewan Cancer Agency; SMS, short message service; TA, treatment adherence.

Table 3.

Clinical studies assessing adherence or treatment discontinuation for oral hormonal therapies in non-mCRPC and beyond.

First author; study type; time range; institutions/databases	Treatment setting; patient eligibility	Treatment regimen(s)	N	Median age (years)	Median follow-up (months) [range]	Measures to improve treatment compliance/adherence	Medication measure	Adherence rate	Discontinuation rate	Other adherence-related measures
Shore et al.14 Phase III, HERO April 2017–October 2018	Advanced PCa Candidates for at least 1 year of ADT	Relugolix	622	72	12	Regular scheduled (biweekly–monthly) and unscheduled patient visits, and phone contacts 1 week prior to each scheduled visit to check on compliance	Pill counting	TAa: >99%	Due to AEs: 3.7%	Completed 6 months of treatment: 90.2% Testosterone recovery (to ⩽ 280 ng/dL) at 90 days: 54%, p = 0.002
		Leuprolide	308	71		NR	NA	TAa: >99%	Due to AEs: 2.6%%	Completed 6 months of treatment: 89.0% Testosterone recovery (to ⩽280 ng/dL) at 90 days: 3%
Pilon et al.53 Retrospective, large prescription claim database analysis October 2014–September 2019 Symphony Health, IDV® database, United States	Advanced PCa Patients initiating apalutamide, enzalutamide, or abiraterone acetate with ⩾6 months of continuous clinical activity prior to and following the index date and diagnosis of PCa	Abiraterone, apalutamide or enzalutamide	27,262	NR	Minimum: 6 Maximum: 18	NR	Prescription orders	@6 months PDC ⩾80%: 57.5% MPR ⩾80%: 92.2% @12 months PDC ⩾80%: 42.8% MPR ⩾80%: 86.1%	NR	@6 months Persistence, No >30 d gap: 53.9% No >60 d gap: 60.1% @12 months Persistence, No >30 d gap: 34.0% No >60 d gap: 40.8%
Grundmark et al.72 Nation-wide population-based study 1997–2006 National Prostate Cancer Register (NPCR) clinical cohort, PCBaSe, linked to the Swedish Prescribed Drug Register, Sweden	PCa Men with a planned first-line monotherapy antiandrogen treatment and without evidence of treatment with GnRH, estrogen, or surgical orchiectomy	Bicalutamide	1406	NR	⩾12	NR	Prescription orders	Median flexible period MPR: 84% MPR ⩾80%: 60%	NR	Median antiandrogen treatment persistence in the localized low/intermediate risk group was almost 6 years, while among men with localized high-risk cancer and regionally metastatic disease, it was slightly more and slightly less than 4 years, respectively. In men with metastatic disease, the median treatment time was 2 years

Studies are ordered by level of adherence evidence.

^aTreatment adherence was defined as the percentage of expected doses actually taken.

ADT, androgen deprivation therapy; AE, adverse events; AR, adherence rate; BAAS, Basel Assessment of Adherence Scale; d, days; GnRH, gonadotropin-releasing hormone; MPR, medication possession ratio; n, number of patients; NA, not applicable; NR, not reported; PCa, prostate cancer; PCBaSE, Prostate Cancer Database Sweden; PDC, prescription days covered; TA, treatment adherence.

Metastatic CRPC

Four prospective^48,51,66,71 and nine retrospective studies^{61–65,67–70} reported adherence outcomes for patients receiving ARATs for mCRPC (Table 2). Three used self-reported measures as the main adherence outcome^66,71 and 10 employed observer-reported measures.^{48,61–65,67–70}

A small prospective, multicenter, open-label, randomized controlled trial (RCT),⁵¹ a large prospective multicenter study⁷¹ and a small multicenter survey study assessed adherence using self-reported measures.⁶⁶ The small RCT assessing abiraterone at two doses (1000 mg fasting and 250 mg with a low-fat meal) showed high initial adherence at 60 days and throughout the study for the lower dose group (n = 19, 96.5% versus 97.9%, p = 0.35) but significantly reduced adherence for the standard higher dose arm at the point in which patients came off study compared to at 60 days (n = 11, 98.2% versus 91.7%, p = 0.008).⁵¹ The large two-armed study conducted by Suttmann et al.⁷¹ assessed the impact of AEMs including education, reminders, and counseling components on adherence assessed with Morisky Medication-Taking Adherence Scale-4 (MMAS-4) scores in patients receiving abiraterone plus prednisone from October 2013 to June 2018.⁷¹ Three-month medium/low MMAS-4 scores were 6.4% versus 9.1% and 6-month scores were 7.1% versus 7.4% among patients completing the MMAS-4 who received AEM (n = 265) versus those not receiving AEM (n = 212) with a 10-month median duration of treatment. Rates of treatment discontinuation not due to disease progression or the start of new cancer therapy were also similar between respective arms [9.0% versus 7.3%, odds ratio (OR) 1.23].

Numerous cohort studies have also assessed adherence using observer-reported measures including nonadherence rates,⁴⁸ proportion of days covered (PDC),^64,67 and medication possession ratios (MPRs)^{61–63,65,68–70} (Table 2). A retrospective national database analysis from January 2012 to December 2013 evaluated abiraterone (n = 5585) and enzalutamide (n = 1413) using prescription orders and at 12 months, similar mean PDCs of 69 and 71% for these agents were demonstrated, respectively, with 47 and 51% of patients showing a PDC ⩾80%.⁶⁷ Discontinuation of study drug occurred in 39 and 33% of patients in the respective groups. A prospective single-institution cohort study by Banna et al.⁴⁸ out of Catania, Italy used pill counting and clinical diaries to assess adherence in mCRPC patients receiving either abiraterone (n = 22) or enzalutamide (n = 36) from April 2014 to December 2018. When adherence was measured at approximately 8.5 months, adherence was high for both abiraterone and enzalutamide, with adherence rates of 96.8 and 93.6%, respectively (p = 0.09) and adherence rates based on the Basel Assessment of Adherence Scale (BAAS) of 99.3 and 98.2% (p = 0.04) for abiraterone and enzalutamide, respectively.⁴⁸

Multiple retrospective registry analyses have assessed adherence in patients with mCRPC using objective measures, including four large studies (over 3000 patients), with 3398–6337 mCRPC patients each.^63,64,68,69 Three demonstrated high median or mean MPRs (92–97%) over variable periods^63,68,69 with two showing 85–95% rates of MPR ⩾ 80% which were comparable for abiraterone and enzalutamide except in the study by Behl et al. (Table 2).^63,68,69 The study by Behl et al.⁶³ also showed a lack of persistence (50.0% for abiraterone and 45.7% for enzalutamide) at 12 months (p = 0.24). The fourth study by Caram et al. showed a 75% rate of PDC ⩾ 80% for abiraterone or enzalutamide and after adjusting for regional-level variables, also found adherence to be higher among patients <70 years of age versus those ⩾85 years of age (76% versus 69%, p < 0.01).^63,64,68,69

The retrospective multicenter registry analysis conducted by Cindolo et al. confirmed findings from three smaller retrospective analyses of registries or administrative databases assessing adherence in mCRPC patients receiving abiraterone or enzalutamide.^61,62,65,70 This study involved mCRPC patients receiving abiraterone (n = 109) and enzalutamide (n = 14) compared to those receiving injections of LHRH analogs (n = 1075) or antagonists (n = 80) in Italy from January 2011 to July 2017.⁶⁵ Median adherence rates as assessed by prescription orders and median persistence, defined as the difference in days of the date of first and last delivery of drug or until the end of available data, were 99.0% and 7.8 months for abiraterone and 95.5% and 8.5 months for enzalutamide, compared with 93.0% and 24.5 months for LHRH analogs and 97.0% and 14.9 months for LHRH antagonists, respectively. This study also investigated the association between adherence and clinical outcomes among the 107 patients classified as super-elderly (9.2%, age range: 85–97 years), showing a median persistence of 13 months and median overall survival (OS) of 29 months among these patients, which was not significantly associated with varying adherence (p = 0.98). OS analysis by co-prescriptions in addition to hormone therapy showed longer OS for patients receiving <5 co-prescriptions (p = 0.005).

Advanced prostate cancer

Data assessing adherence for oral hormonal therapy in advanced prostate cancer, including the castrate-sensitive and nonmetastatic settings, are just now beginning to emerge (Table 3). At present, there are data from one RCT assessing ADT and two retrospective cohort studies reporting outcomes for androgen receptor antagonists.^14,53,72 The phase III HERO study conducted by Shore et al.¹⁴ randomized 930 patients with high-risk advanced prostate cancer. Eligible participants, of whom 12% had received prior ADT, were randomized to receive at least 1 year of continuous ADT 2:1 with either the oral LHRH antagonist relugolix (n = 622) or the LHRH agonist leuprolide by injection (n = 308).¹⁴ Adherence was assessed through scheduled or unscheduled visits and phone calls and adherence was evaluated through pill counting. At a median follow-up of 12 months, treatment adherence, defined as the percentage of scheduled doses taken, was >99% in both arms, while rates of treatment completion at 48 months were 90.2% versus 89.0%, for relugolix versus leuprolide, respectively. Treatment discontinuation due to adverse events (AEs) was not reported.

The next largest cohort study was a nation-wide population-based study conducted by Grundmark et al.⁷² in Sweden from 1 January 1997 to 31 December 2006. The study assessed adherence to the ARAT bicalutamide among 1406 prostate cancer patients who had a planned first-line ADT monotherapy and no evidence of treatment with gonadotropin-releasing hormone (GnRH), estrogen, or surgical orchiectomy using prescription fill orders to determine MPRs.⁷² With at least 12 months of follow-up, the median MPR was 84% and MPR > 80% were 60.4%. Treatment discontinuation due to AEs was not reported. Finally, a retrospective analysis of a large prescription claim database assessed apalutamide, enzalutamide, or abiraterone from October 2014 to September 2019 using prescription orders of 27,262 patients with advanced prostate cancer who had shown ⩾6 months of continuous clinical activity prior to and following the first approved claim.⁵³ Data showed an MPR ⩾ 80% of 92.2% at 6 months and 86.1% at 12 months as well as a PDC ⩾ 80% of 57.5% at 6 months and 42.8% at 12 months. Persistence, defined as patients with no >60-day gap in treatment between adjacent fills, was 60.1% at 6 months and 40.8% at 12 months.

Discussion

Oral hormonal therapies, ARATs and ADT, are an integral part of the treatment of advanced prostate cancer, and maintaining adherence rates to oral therapies has been key to ensuring optimal outcomes in cancer. Clinical trial and real-world data on adherence to oral hormonal agents used in prostate cancer were summarized, of which the majority were real-world, observational studies (94%, n = 15) conducted in mCRPC (81%, n = 13).^{48,51,53,61–72} There was only one phase III trial prospectively evaluating adherence in advanced prostate cancer.¹⁴

How do adherence outcomes for oral hormonal therapy change based on type of measure used and duration of follow-up in advanced prostate cancer?

The majority of studies reviewed primarily used MPR-based (50%)^{61–63,65,68–70,72} or adherence rate-based (19%) endpoints,^14,48,66 which may not fully account for discontinuation (Figure 2(a)). PDC-based endpoints, which do factor in discontinuation, were less common (19%)^53,64,67 followed by studies using questionnaire-based indices as the main adherence measure (12%).^51,71 In addition, when reported, follow-up time ranged from 6 to 12 months, with most studies reporting outcomes between >9 and 12 months (56%, Figure 2(b)).

Figure 2.

Distribution of (a) main adherence measure and (b) follow-up time in clinical studies assessing adherence to oral hormonal therapies in advanced prostate cancer.

When exact adherence follow-up (i.e. adherence at X months) was not available, mean or median treatment duration was used. Studies looking at adherence over follow-up periods defined as ‘at least X months’ were considered to have follow-up of the minimal cutoff used, that is, ‘X’. Follow-up in studies with different follow-up periods for different agents was calculated as the average follow-up time across agent cohorts.

ACI, Adherence Composite Index; AR, adherence rate; MMAS-4, Morisky Medication-Taking Adherence Scale-4 item score; mos, months; MPR, medication possession ratio; NR, not reported; PDC, prescription days covered.

When studies reporting adherence measures closest to 12 months of follow-up were considered, observer-reported studies reporting total dose rates which measure adherence through pill counting (adherence rate, >93.6%),^14,48 claim data mean (>92%), and median MPR rates (>99.5%)^63,69,70 were generally high and comparable for abiraterone and enzalutamide (Figure 3(a)). However, mean coverage rates (>69%),⁶⁷ patient ratios based on MPR (MPR⩾80%; >84.6%)^63,70 or PDC (PDC⩾80%; >42.8%),^53,67 or rates of persistence (>45.7%)^53,63 were generally lower with PDC ratios being higher for enzalutamide and persistence rates being higher for abiraterone. When self-reported measures were considered, adherence rates ranging from 99.3 to 92.6% were seen.^48,51,71

Figure 3.

Values for the different adherence measures reported across different studies (a) at 12 months or at the closest follow-up time and (b) at different timepoints.

ACI, Adherence Composite Index; AR, adherence rate; ARATs, androgen receptor-axis-targeted therapies; BAAS, Basel Assessment of Adherence Scale; DoT, duration of treatment; MMAS-4, Morisky Medication-Taking Adherence Scale-4 item score; mos, months; MPR, medication possession ratio; PDC, prescription days covered.

The Behl and Pilon studies assessed adherence outcomes over multiple time periods and therefore provide particular insight into how adherence outcomes change over time (Figure 3(b)).^53,63 In Behl et al., modest drops in either MPR dose ratios (>97 to >92%, 3–12 months, respectively) or MPR patient ratios (>93.9 to >85.4%, 3–12 months, respectively) were seen over time. However, more dramatic differences in persistence rates were observed over time with persistence rates dropping from >91.1% at 3 months to >45.7% at 12 months in the Behl et al. study and from 60.1% at 6 months to 40.8% at 12 months in the Pilon et al. study. There is a need for more studies assessing adherence to oral hormonal therapy in advanced prostate cancer which take into account both dose and patient ratios and follow patients beyond 12 months.

Does poor adherence affect clinical outcomes in advanced prostate cancer?

Maintaining castrate levels of testosterone in advanced prostate cancer is a key therapeutic goal and has been linked to improved clinical outcomes.^6,13,56–58 Although low adherence rates have not been linked to low levels of testosterone, it would be reasonable to assume that poor adherence rates could lead to testosterone breakthroughs which have been associated with compromised outcomes in larger studies.⁵⁹ This is especially concerning with agents that require continued, daily intake as nonadherence may lead to transient testosterone elevations that may remain undetected between testosterone checkups. In early breast cancer, low rates of adherence to oral hormonal therapy have been linked to reduced survival and potentially quality of life.^73–80 While the retrospective analysis by Cindolo et al.⁶⁵ failed to demonstrate an association between OS and adherence rates (p = 0.98), Banna et al.⁴⁸ found an association between nonadherence and radiological response by receiver operating characteristic curve analysis but not with radiological progression-free survival or OS. A strong correlation between OS and duration of abiraterone treatment was identified by Al-Ali et al.⁶² in their retrospective analysis of the claims database from the largest insurance company in Austria. However, duration of treatment in the mCRPC setting may be predominantly determined by disease progression events (rather than treatment discontinuation due to noncompliance) and is therefore unclear the impact of treatment compliance on the reported correlation with OS. Although there is no long-term real-world adherence data on use of oral ADT, it is reasonable to expect that low adherence rates could negatively impact clinical outcomes in this setting.

What factors affect adherence to oral hormonal therapy?

Multiple factors can negatively affect adherence to oral hormonal therapy including patient-related, treatment-related, and schedule-related factors (Table 4).^{73,77,80–85} Significant associations with nonadherence were not consistently found across studies, which may be explained by differences in (i) study and effect size and statistical power to detect significant associations; (ii) adherence measures used (more difficult to detect associations with less sensitive measures); (iii) scales and thresholds used to assess factors; (iv) statistical methods (Spearman rank correlation, covariate cluster analysis, univariable and multivariable linear or logistic regression analysis or Cox proportional hazards models, and stratified analysis); and (v) the presence of confounding variables, among others. Given the observational and retrospective nature of most studies and analyses, it is difficult to draw any firm conclusions regarding the predictive value of factors tested, particularly when associations are inconsistent or not found. Therefore, the associations identified across studies can only be considered hypothesis generating at this time.

Table 4.

Factors associated with nonadherence in clinical studies assessing adherence to oral hormonal therapies in advanced prostate cancer.

Variables	Association with nonadherence or related endpoints
Patient-related and socioeconomic
Older age and frailty	⩾75 years (versus <65 or <75 years)53,72; ⩾85 years (versus <70 years)64; geriatric G8 score ⩽1448
	Non-persistence: >85 years (versus ⩽85 years)65
	Shorter time on treatment: >75 years (versus ⩽75 years)68
	No significant association 48,51,69
Multiple comorbidities	Per unit increase in Quan-CCI53
	Shorter time on treatment: High CCI (⩾3)68
	No significant association: CCI 48,72
Low-income or poor socioeconomic status	Low-income subsidies64
	No significant association: Education, employment before illness, employment currently and income51; socioeconomic status tertile64; socioeconomic status72
Race	Black patients (versus White patients)53; living in an HRR with a higher proportion of Hispanic patients64
	No significant association: Ethnicity51; African-American or Hispanic (versus White) and living in an HRR with higher proportion of African-American patients64; Hispanic patients (versus White patients)53
Remote location. Geographic region	Non-persistence: South (versus Midwest United States)53
	No significant association: West and Northeast (versus South United States) 53
Patient’s beliefs about effectiveness of therapy. Forgetfulness	Misperception on the need of the drug and forgetfulness (most frequent patient-reported reasons for nonadherence)48
Treatment related or disease related
Disease stage	Low-risk localized (versus high-risk localized and metastatic)72
	No significant association: PSA at treatment start48; the absence of metastases (nearly significant)53
Prior and concurrent treatment	CT use in the 6 months prior to index prescription53; no first-generation antiandrogen use53 Shorter time on treatment: Prior CT, short time on prior ADT and non-curative primary treatment68
	No significant association: Setting of therapy (pre-CT versus post-CT)48; recent CT69; steroid use and concomitant drugs48
Treatment intolerance or toxicity	High rates of single fills and low number of prescription fills67
	No significant association: G1–2 and G3–4 toxicity 48
Lack of response to therapy or low symptom control	No radiological response48; worse symptom control66
	No significant association: PSA response and −ΔPSA 48
Treatment schedule related
Dose	Increasing dose51
Overall complexity of drug regimens	Four tablets fasting (versus one tablet with low-fat meal)51
Duration	No significant association: Treatment duration (no. of cycles) 48
Cost	Insured by commercial plan (versus Medicare)53; living in an HRR with lower proportion64 of Medicaid-eligible patientsb; primary insurer and out-of-pocket paid amount above the median (versus below the median)53
	No significant association: Medicaid versus commercial healthcare plan 53,69

Factors belonging to relevant patient-, disease-, or treatment-related categories were included in the table if their association with nonadherence was reported in eligible studies of adherence to oral hormonal therapy in prostate cancer. For each factor, positive associations (statistically significant or reported as positive trend) are listed at the top, while negative (statistically nonsignificant) associations at the bottom in italicized font. Factors and comparisons are presented to show those associated with lower adherence in statistically significant associations. Additional factors were assessed.

^aAdditional factors for which no association was found were as follows: caregiver presence,⁴⁸ marital status,^51,72 and first prescriber (medical specialty of physician initiating the treatment).⁷² A significant association between adherence and index year was found by Pilon et al.⁵³ (lower adherence in 2016–2019 versus 2015).

^bReported trend toward significance (0.05 <p< 0.1).

ADT, androgen deprivation therapy; CCI, Charlson Comorbidity Index; CT, chemotherapy; G, grade; HRR, hospital referral region; PSA, prostate-specific antigen.

The most common patient-related factor associated with nonadherence was older age,^53,64,68,72 which is often associated with low income,⁶⁴ frailty, forgetfulness, and multiple comorbidities, factors also found to negatively affect adherence.^48,65,68 Other patient-related factors negatively affecting adherence to ARATs in advanced prostate cancer were race, poor socioeconomic status, and remote location.^53,64 Although some suspect that male gender may be associated with poor adherence, inconsistent associations between gender and adherence have been observed in studies of hypertension.⁸⁶ The most common treatment-related factors negatively affecting adherence were earlier stage of disease (better prognosis),⁷² treatment intolerance,⁶⁷ concomitant medication,⁵³ and lack of response to therapy or low symptom control,^48,66 although patients’ beliefs about ineffectiveness of therapy also had an impact. The most common schedule-related factors were increasing dose (number of tablets per day),⁵¹ overall complexity of drug regimens, and higher cost.^51,53 Administration frequency has also been identified as a factor negatively impacting adherence to injectable LHRH agonists with shorter intervals associated with lower adherence.¹⁰ Similar factors have been associated with poor adherence to oral hormonal therapy in breast cancer^{77,84,85,87,88} and oral anticancer therapy in general.^{37,38,41,89,90} Adverse events, lower socioeconomic status, higher medication cost, lack of belief in efficacy/necessity of treatment, and perception of lower risk of recurrence^{77,84,85,87,88} might be particularly impactful contributors to nonadherence in early settings where disease severity and urgency of treatment are not as apparent for patients.

What measures can be taken to improve adherence to oral hormonal therapy?

Strategies to improve adherence can include the identification of patients at risk of poor adherence and removing barriers to adherence, in addition to AEM to provide support.^71,90,91 Removing barriers to adherence can include reducing treatment complexity by eliminating unneeded medications, adjusting medicines for minimizing contraindications and improving tolerance, and increasing the interval between doses by favoring long-acting formulations. Adherence can also be improved by eliminating unnecessary medications and simplifying schedules (e.g. reduced number of doses per day). Providing financial assistance to some patients may also improve adherence, although this is less likely a factor in single-payer jurisdictions. AEMs can include educational tools, counseling, and reminders in the form of study nurse calls, patient diaries, and text messaging services.⁷¹ Only Suttmann et al.⁷¹ evaluating abiraterone for mCRPC explored the impact of AEMs and found no benefit. Although patients in this study were older with multiple comorbidities, a population that should benefit from this approach, 40.8% did not regularly utilize the measures, adherence overall was high and discontinuation rates were low (8.3%) which could have attenuated adherence-enhancement benefit. Nevertheless, De la Llave et al.⁶⁶ reported high and consistent adherence rates to androgen receptor inhibitors despite low utilization of devices to enhance adherence and several systematic reviews found minimal impact for educational and reminder AEMs on oral cancer therapies.^{40,71,90–92} Systematic reviews covering multiple settings found that effects of AEMs were inconsistent, and improvements only observed in a minority of studies.^93,94 However, a systematic review on mobile health interventions for patients with cancer noted that interventions associated with statistical improvements in adherence included interactive, multidimensional components that allowed for feedback and communication with medical providers, educational tools, and/or social networking.⁹⁵

What are the implications of use of LHRH inhibitors in advanced prostate cancer?

As many prostate cancer patients will have prolonged treatment duration, are of advanced age, and have multiple co-morbidities, adherence to LHRH inhibitors is likely to be an issue. Treatments that are tolerable in addition to effective are therefore preferred to ensure adherence. The phase III HERO study comparing oral and injectable LHRH inhibitors found both to be tolerable and effective with good adherence at 1 year. However, this study provided little insight into long-term adherence or persistence and may not be transferable to real-world experience.

A retrospective study by Cindolo et al.⁶⁵ provided additional insight into oral versus injectable hormonal therapy, while also providing relative persistence data in patients with mCRPC who started on LHRH analogs, LHRH antagonists, enzalutamide, and abiraterone.⁶⁵ Median MPR rates were high for all agents, although slightly lower for LHRH analogs (93%) compared with LHRH antagonists, abiraterone (99%), and enzalutamide (95.5%). However, median persistence was nearly twice as long for LHRH analogs (24.5 months) compared with LHRH antagonists (14.9 months) and three times as long compared with both androgen receptor inhibitors (7.8–8.5 months), suggesting that short-term adherence may not be a good indicator of long-term persistence. Furthermore, this study showed no association between adherence rates and survival (p = 0.98), also suggesting that traditional adherence measurements may not provide the full picture. A retrospective study by George et al.¹⁰ using data from 12,843 patients examined adherence to primary and secondary LHRH agonists for up to 3 years.¹⁰ Rates of MPR >80% were 79 and 71%, respectively, and adherence was threefold higher with 180- to 365-day injection intervals compared with 90-day intervals (ORs: 2.60–2.65), with the greatest adherence seen for patients given the 365-day injection implant compared with 90-day injection intervals (OR 3.29). Oral and injectable LHRH inhibitors are effective and safe, and long-lasting injectable formulations allowing increased treatment intervals will likely translate into improved adherence.

Conclusion

Prostate cancer patients are at a high risk of poor adherence, which has been associated with compromised clinical outcomes. Here, data on adherence to oral hormonal therapies in advanced prostate cancer have been identified and summarized. Evidence is generally of low quality, with high heterogeneity and inconsistent reporting across studies. The majority of adherence data are in the metastatic castrate-resistant setting, with less data on adherence in the nonmetastatic and hormone sensitive settings where patients may undergo prolonged therapy. Although possession ratios assessed up to a year of therapy are high in the mCRPC setting, coverage and persistence rates are lower and decline after 6 months. As poor adherence could be linked to compromised outcomes, particularly in patients who are facing decades of therapy, there is a need for additional studies using sensitive measures that capture different aspects of adherence over years and across healthcare systems as well as a need to better understand adherence risk factors. Finally, it is important to identify patients at risk of poor adherence while minimizing obvious barriers to adherence such as optimizing dose schedule.