Factors Associated with Suboptimal Adherence to Tyrosine Kinase Inhibitors in Patients with Renal Cell Carcinoma—A Retrospective Cohort Study

Abstract

Background: Adherence to tyrosine kinase inhibitors (TKIs), the first-line treatment for renal cell carcinoma (RCC), is critical to ensure intended treatment outcomes. However, 75% of patients with RCC have persistency gaps (>7 days) within the first 90 days after initiating TKIs. This study explored factors affecting TKI adherence in RCC patients to inform future interventions. Methods: A retrospective cohort study was conducted at a specialist oncology hospital in Northwest England from October 2020 to October 2022 on patients with RCC treated with TKIs. TKI prescriptions and persistence gaps (>7 days) were identified from electronic dispensing records. Factors associated with persistence gaps were retrieved by reviewing patients’ clinical records. We used descriptive statistics to summarise the results and Kaplan–Meier analysis to assess the probability and the time to the first gap, stratified by adverse drug effect (ADE)-related and non-ADE-related gaps. Results: Among 165 included patients, 611 persistence gaps were identified. ADEs accounted for 59% (n = 464) of 787 recorded factors, with diarrhoea being the most frequent ADE (9.5%). Patients holding leftover TKIs were the primary (15.1%) non-ADE factor for persistency gaps. At least one gap was observed with 82% of patients (n = 135); 19% had ≥5 ADE-related gaps, and 25% had ≥5 non-ADE-related gaps. ADE-related gaps typically occurred within the first three months (50%), while non-ADE-related gaps were not time-dependent. Conclusions: ADEs, including diarrhoea and pain-related reactions, were the most frequently reported issues affecting TKI persistency in patients with RCC. These ADEs are likely to impact patients’ quality of life and adherence. Future qualitative research is warranted to explore patients’ care needs and additional factors such as health literacy and self-efficacy.

1. Introduction

Renal cell carcinoma (RCC) is the most common type of kidney cancer, with 80% of kidney cancers classified as such [1]. Globally, RCC is the sixth most commonly diagnosed cancer in men and the 10th in women [2]. The incidence is higher in men than in women, with a relative risk of 1.7, and this pattern is consistent across all age groups [2,3]. The highest incidence rates are observed in North America, Western Europe, and Australia/New Zealand, with Age-Standardised Rates (ASRs) of 10.9, 9.7 and 9.6 per 100,000, respectively [3].

Concerningly, the incidence of kidney cancer has increased significantly over the past three decades, with the global ASR rising from 3.5 per 100,000 in 1990 to 4.6 per 100,000 in 2019—a 29.1% increase [4]. This rise is attributed to global ageing, increased life expectancy and improved tumour detection and reporting [4,5]. In the United Kingdom (UK), kidney cancer is the 7th most common cancer, with an 88% increase in incidence since the early 90s, compared to 18% for breast cancer and 9% for lung cancer [6,7]. This underscores the need for adequate resource allocation to address the growing RCC incidence [4].

Traditional oncology therapies like radiation and chemotherapy are ineffective in treating RCC [8]. For around 20 years, from the late 1980s until the mid-late 2000s, immunotherapy, such as interleukin-2 and interferon therapy, was the mainstay of RCC treatment, but these had low response rates and high toxicity [9,10,11]. Investigation into angiogenesis, crucial for tumour growth in RCC, led to the development of tyrosine kinase inhibitors (TKIs) targeting receptors for vascular endothelial growth factor (VEGF), revolutionising RCC treatment [12]. Sunitinib, introduced in 2007, became the first-line treatment for RCC, followed by other TKIs and combination therapies with immunotherapy [13].

Adherence to prescribed medication is critical for the efficacy of these life-extending treatments [14]. Adherence involves therapy initiation, implementation, and discontinuation, with persistence measuring the duration from initiation to the last dose before discontinuation [14]. There is no gold standard for measuring adherence, and a combination of measures is considered best practice [15], with direct methods like blood assays being optimal but invasive and indirect methods like dispensing data analysis being less resource intensive [16]. Persistence, a proxy for adherence, is often calculated using the number of days therapy was available within a “permissible gap” between refills [17].

Contrary to expectations [18], real-world data show poor adherence and persistence for oral systemic anticancer therapies (SACTs), including TKIs for RCC [19]. Real-world data is collected outside of conventional randomised controlled trials (RCTs) and reflects routine clinical practice [20]. These data are considered more representative of the broader patient population being studied and therefore offer greater generalisability.

Clinical trials report high persistence rates, such as 91% at 19 months for imatinib in Chronic Myeloid Leukaemia (CML) [21], but real-world adherence is much lower, ranging from 26.4% to 36.1% [22,23]. As seen with imatinib in CML, poor adherence reduced molecular response levels and survival rates [22]. Angus et al. (2024) analysed refill data from 109 RCC outpatients in Northwest England. They found that 75% of RCC patients had at least one persistence gap, defined as a gap of more than seven days between the actual and planned refill dates within the first 90 days of treatment, and 92% of patients had a gap within the first 180 days of treatment [24]. However, factors associated with and the impacts of suboptimal adherence to TKIs in RCC patients remain understudied.

The World Health Organisation (WHO) identifies five factors affecting adherence: socioeconomic and therapy-, health-system-, condition-, and patient-related factors. Effective interventions must address all these dimensions [25]. RCC patients tend to be older and face unique challenges when taking TKIs, including adverse effects, cancer symptoms, and polypharmacy due to comorbidities, which exacerbate adherence issues. Therefore, understanding the issues and challenges of suboptimal adherence will help design future interventions to support cancer survivorship. This study investigated potential factors associated with TKI persistency gaps, a proxy of suboptimal adherence, in patients with RCC cancer to explore the future direction of specialised adherence-aiding approaches to improve patient care.

2. Results

2.1. Patient Characteristics

Three hundred sixty-four patients were identified as receiving TKIs during the cohort inclusion period (December 2020 to December 2022). Of these, 165 patients met the inclusion criteria with a median age of 73 years (Interquartile Range (IQR): 64, 77). Males comprised 66% of the cohort. The median follow-up was 34.8 months (IQR: 18.1, 54.7). Approximately 65% were followed for over two years, and 10% for less than one year. At the end of the follow-up, 43 patients were still alive.

2.2. Tyrosine Kinase Inhibitor Use and Persistency Gaps

From 5093 dispensing records for 165 patients, sunitinib (33.9%) and cabozantinib (22.4%) were the most prescribed TKIs. Over half of the patients (n = 88, 53.3%) switched to a second TKI, with the most frequent cabozantinib (53.4%). A small group (n = 11, 6.7%) received a third TKI, primarily cabozantinib (45.5%) (Figure S1).

Overall, 611 persistency gaps (≥7 days) were identified after data cleaning. Among the 165 patients, 82% had at least one gap, and 39% had five or more (Figure 1). The median gap length was 15 days (IQR: 10, 28). Half of the patients (50%) experienced their first gap within four months of treatment initiation, increasing to 75% within the first year, and only 5% had their first gap in the second year (Figure 2).

2.3. Factors Associated with the Persistency Gaps

A total of 787 factors potentially associated with 611 persistency gaps were identified from the electronic medical records (EMRs). Each gap could be linked to multiple factors. Of these, 464 (59%) were related to Adverse Drug Events (ADEs) and 171 (21%) were associated with non-clinical factors (

Table 1. Factors associated with persistency gaps of tyrosine kinase inhibitors in patients with renal cell carcinoma.

Category	Correlated Factors	Count	Proportion (%) *
General clinical factors	Hospital admission	47	6.0%
	Drug changes	33	4.2%
	Operation	29	3.7%
	COVID infection	9	1.1%
		118	15.0%
Non-clinical factors	Holidays	21	2.7%
	Sufficient leftover	119	15.1%
	Loss of contact or missed delivery	31	3.9%
		171	21.7%
Adverse drug events	Diarrhoea	75	9.5%
	Neutropenia or infection risk	47	6.0%
	Pain	47	6.0%
	Fatigue or other central nervous system disturbance	43	5.5%
	Mucositis	36	4.6%
	Nausea or vomiting	33	4.2%
	Hypertension	28	3.6%
	Liver toxicity	18	2.3%
	Dysgeusia	17	2.2%
	Palmar Plantar Erythrodysesthesia	17	2.2%
	Impaired wound healing	16	2.0%
	Other skin reaction	16	2.0%
	Poor appetite	11	1.4%
	Other adverse events	60	7.6%
		464	59.0%
Others		34	4.3%

* The proportion of the factor contributes to all 787 records of factors.

).

The top five most frequent ADEs were diarrhoea (9.5%), neutropenia or infection risk (6%), pain (6%), fatigue or other central nervous system disturbance (5.5%) and mucositis (4.6%). The severity of ADEs varied, with less severe events like diarrhoea (9.5%) and palmar-plantar erythrodysesthesia (2.2%) differing from more severe ADEs such as hypertension (3.6%) and mucositis (4.6%) (Table 1).

Among non-ADE factors, “holding sufficient leftover TKIs” was the most common factor (15.1%). General clinical factors such as switching to immunotherapy (4.2%), hospital admissions (6%), and operations (4.2%) were also significant. Hospital admissions excluded admissions for operations but included reasons such as hospitalisation for infection (Table 1).

2.4. Time to the First Adverse Drug Event Associated Persistency Gap

When stratifying all 165 patients by whether their first persistency gap was linked to an ADE or a non-ADE, the Kaplan–Meier curve on the proportion of patients without a persistency gap over time showed that 50% of the ADE group experienced a persistency gap within three months, with the decreasing proportion plateaued at the 45th month (Figure 3).

Conversely, the non-ADE group showed a gradual, linear decline trend in the proportion of patients without a persistency gap over time after the 6th month, indicating a consistent rate of non-ADEs.

Further analysis revealed that 40% (n = 46) of patients with an ADE-related first gap had no subsequent gaps. However, among patients who had already experienced persistence gaps, those with three or four gaps were proportionally more likely to go on to have additional gaps than those with only one or two gaps.

A fair proportion of the patients who experienced at least one gap went on to experience five or more gaps (ADE: n = 22, 19%; non-ADE: n = 26, 25%), as shown in Supplementary Figure S2.

3. Discussion

This single-centre retrospective cohort study identified suboptimal adherence among patients with RCC, as evidenced by gaps in prescription refills. Specifically, 82% of patients experienced at least one gap in their TKI regimen, with 39% having five or more gaps. These findings align with a previous retrospective study at the same hospital (November 2021 to March 2022), where 109 RCC patients on TKIs were followed for an average of 2.6 ± 1.4 years [24]. The median Medication Possession Ratio (MPR) was 95.6% (IQR: 90.7–100.1%), with 75% and 92% of patients experiencing a persistency gap of ≥7 days within the first 90 and 180 days, respectively [24].

Comparing our findings with those of other studies is challenging due to the limited research on persistency gaps as an adherence measure for SACT in patients with RCC. Only two other studies [26,27] reported MPR of TKIs in patients with RCC, where patients’ mean MPR of sunitinib and pazopanib as first-line treatment was 0.76 ± 0.19 and 0.86 ± 0.19, respectively [27]. However, differences in methodologies and patient populations in these studies limit direct comparability with our findings, as they were based in the United States, where the health insurance system plays a crucial role in treatment accessibility.

Building on our previous research, this study identified adverse drug events as the leading cause of persistency gaps in TKIs among RCC patients, accounting for 59% of documented factors. This finding aligns with existing evidence that side effects from targeted therapies often reduce adherence. For example, a survey found that, despite the higher incidence of side effects with targeted therapies, adherence rates were similar to those with hormone therapy plus chemotherapy [26]. Similarly, a qualitative study in Malaysia reported that TKI-related ADEs significantly impacted adherence in patients with CML [27].

This study further revealed that ADE-related medication gaps most frequently occur within the first three months of treatment, with a 50% incidence rate. This result contrasts with the CLEAR study, which reported a median time to the first critical ADE at five months for patients on lenvatinib plus pembrolizumab versus sunitinib [28]. The discrepancy may stem from the immediate and pronounced effects of TKIs, leading to earlier onset of toxicities. Moreover, our cohort’s limited use of lenvatinib compared to the CLEAR study may also explain differences in ADE timing and occurrence.

The study’s findings highlight the importance of focusing on adherence interventions during the first three months of treatment, which are marked by frequent adherence gaps. This critical window offers an opportunity to implement strategies to improve adherence and manage adverse effects. Specifically, attention should be given to managing severe adverse events, such as diarrhoea, reported as the most frequent ADE. Developing patient-centred support services that offer targeted advice on managing diarrhoea, including rehydration and absorption issues with other oral medications [29] could significantly enhance adherence and treatment outcomes.

Several ADE categories identified in this study are associated with pain, including palmar-plantar erythrodysesthesia, mucositis, fatigue, and pain as a standalone category. Upon further examination, the cumulative incidence of pain-related ADEs (18.3%) exceeds that of diarrhoea (9.5%), underscoring the clinical relevance of pain in this patient population. Prior research has demonstrated that pain-related ADEs can substantially impair patients’ quality of life and contribute to non-adherence to tyrosine kinase inhibitor (TKI) therapy [30]. These findings reinforce the need for qualitative research to explore patients’ lived experiences and perceptions of ADEs, which will inform patient-centred care strategies and adherence support.

Moreover, further research is essential to develop effective adherence interventions. Qualitative approaches, such as focus groups capturing patients’ perspectives, values, and their impact on adherence, treatment, and quality of life, will offer valuable insights into the patient experience and factors affecting adherence. Additionally, conducting large-scale, multicenter quantitative prospective cohort studies is crucial to explore a broader range of covariates to validate current findings and provide a more comprehensive understanding of adherence challenges. This research approach will strengthen the evidence base in this under-explored area and guide the design of targeted interventions to improve adherence and treatment outcomes.

A pharmacist-led service could significantly optimise patient care, particularly in managing medications and associated adverse events. For instance, a study in Malaysia found that pharmacist-led consultations addressed adverse events (45.5%) and treatment ineffectiveness (31.5%), with 83.7% of 233 interventions being implemented and 77.3% of issues resolved [31]. However, due to differences between Malaysian and the UK healthcare systems, disease profiles and cultural backgrounds, it is essential to explore whether similar outcomes could be achieved within the UK National Health Service (NHS) for RCC treatment. Conducting such studies in the UK would provide valuable insights into the feasibility and effectiveness of these services.

This study is among the few that focus on adherence to VEGF-TKI therapies in RCC, particularly the impact of adverse events. It provides valuable insights into the challenges of managing RCC treatment and highlights critical factors influencing adherence. However, it also identifies knowledge gaps, such as the roles of forgetfulness, polypharmacy, and supportive medications, which were not empirically examined. Addressing these gaps in future research could enhance understanding and improve adherence-support strategies.

This study used persistency gaps to measure adherence instead of MPR. This choice addresses the limitations and lack of a universally accepted gold standard for adherence measures [15,16]. Persistency gaps benefit studies with varying follow-up lengths, enabling comparison across different durations without introducing significant bias. MPR can be misleading in such cases, as shorter follow-up periods often result in higher MPR values, potentially inflating adherence rates due to reduced opportunities for gaps [32]. This study offers a more accurate and unbiased assessment of adherence across varying follow-up intervals by using persistency gaps.

This study also benefits from utilising electronic prescriptions and clinical records, providing a non-invasive, anonymous data collection method that minimises ethical privacy concerns. This approach enables the investigation of a large patient population over extended periods, yielding a robust dataset for analysis. Electronic records support comprehensive longitudinal studies, enhancing the reliability and generalisability of findings across diverse patient groups and long follow-up durations [16].

Although using persistency gaps to measure adherence provides valuable insights, it has limitations. The 7-day cut-off for defining a persistency gap was based on TKIs’ half-lives, pharmacists’ advice, and prior studies [24]. However, this cut-off may not apply to all TKIs due to differences in their pharmacological profiles. For example, a seven-day gap could have varying clinical implications for patients on pazopanib (half-life: 30.9 h) versus those on cabozantinib (half-life: 110 h) [33]. Future research is needed to establish a universal standard for persistency gaps that considers the pharmacokinetic characteristics of different TKIs.

The terminology surrounding ADEs and Adverse Drugs Reactions (ADRs) remains inconsistent across pharmacoepidemiologic and pharmacovigilance literature [34]. According to WHO, ADEs refer to any untoward medical occurrences that may arise during medication use, without necessarily implying a causal relationship. In contrast, ADRs are defined as harmful or unpleasant reactions resulting from the use of medicinal products at normal doses, where a causal link is either established or strongly suspected [35]. ADRs therefore represent a more specific subset of ADEs, with clearer clinical relevance due to the implied causality.

Nonetheless, several studies have adopted the broader definition of ADEs to encompass drug-related injuries, including those with varying degrees of causality [36]. Furthermore, the terms ADE and ADR are frequently used interchangeably in the literature [37]. Given this prevailing usage and the scope of our study, we have retained the term “ADEs” while acknowledging its limitations and the importance of causality in clinical interpretation.

The data source used to identify factors associated with persistency gaps was the EMRs. The researcher reviewed clinicians’ notes to determine whether a factor was likely to be the cause of a persistence gap, based on clinical context and timing. Only factors judged to be causative were recorded; those considered consequences of a gap were excluded. We acknowledge that this approach may introduce observer bias, as a single reviewer conducted the assessment. Future studies should consider involving multiple reviewers to enhance reliability and reduce subjectivity.

Another limitation is the potential for researchers to interpret clinical notes subjectively. Clear criteria should be established for identifying correlating factors in clinical notes to improve reliability in future studies. Additionally, the reliance on patients’ self-reported medication histories risks inaccuracies due to potential over-reporting or under-reporting [38].

Inclusion of polypharmacy and co-morbidities of patients would have enabled a more comprehensive analysis of patient characteristics and treatment barriers, as knowledge of these factors would aid a deeper understanding of adherence behaviour. Comparing the prevalence of factors between individuals with and without a persistence gap in future studies would offer enhanced understanding of which factors most strongly influence persistence gaps.

Only people using TKI as monotherapy were included in this study, as including patients on combination therapies would introduce additional confounding variables that could affect adherence. This approach allows the findings to more accurately reflect factors associated with TKI use alone, rather than being influenced by concurrent treatments such as immunotherapy or everolimus.

The study also encountered challenges with missing data: 16% of gaps lacked an associated factor, and 5% were erroneous upon reviewing clinical notes. These issues may result from missing variables or data collection errors. Using a single data collector without a second assessor for verification could have introduced inaccuracies and biases. A dual-review process with a second researcher could enhance data accuracy and minimise overlooked errors.

The current treatment landscape for RCC often involves a combination of targeted therapies and immunotherapy in first-line treatment [39]. In recent years, immune checkpoint inhibitors like nivolumab, avelumab and pembrolizumab have also become key components of RCC therapy, helping the immune system recognise and attack cancer cells. Combining TKIs with immunotherapy has shown improved outcomes in RCC patients, offering a more effective approach than either treatment alone [40]. This study examines the importance of adherence to oral TKIs when used as monotherapy, as adherence to oral treatment is a key component of combination therapy in RCC.

4. Materials and Methods

4.1. Study Design and Setting

This retrospective cohort study was conducted at a leading specialist oncology centre in Northwest England, serving approximately 60,000 patients annually from a population of 3.2 million across Greater Manchester and Cheshire [41]. The data collection period was from November 2022 to March 2023, using hospital dispensing and electronic medical records.

Ethical approval was not required as it was regarded as a service evaluation approved by the hospital’s Quality Improvement and Clinical Audit department. Several site visits were conducted to shadow and interview the renal cancer team, ensuring the study’s aims, objectives, and design were relevant and feasible (Figure 4). In addition, this study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement guidelines [42] (Table S1).

4.2. Data Source

Several data sources were used for this study (Table S2). The cyclical dispensing scheme (CDS) spreadsheet was used to identify the study cohort and their prescribed SACT. The CDS, a pharmacy technician-led system, dispenses monthly SACT prescriptions to stable cancer patients between consultations, usually every three months [43]. Iqemo, an electronic SACT management system, provided demographic information and SACT indications. The JAC dispensing system recorded the date and amount of each SACT dispensing. The Christie Web Portal (CWP) contained electronic medical records (EMRs) from healthcare professionals, which were used to explain the reasons for identified persistency gaps.

4.3. Study Cohort and Selection

The study included adults (>18 years) with advanced RCC enrolled in the CDS between December 2020 and December 2022 who met the inclusion criteria. Patients were prescribed single-agent TKIs (sunitinib, axitinib, pazopanib, tivozanib, or cabozantinib) approved by the National Institute for Health and Care Excellence for advanced RCC [13]. Exclusions were for TKIs prescribed through compassionate use schemes, combined therapies with immunotherapy or other SACT, and lenvatinib, which is not used as monotherapy, only combined with everolimus. Eligible patients identified were pseudonymised, and the list of hospital numbers alongside pseudo numbers was stored on encrypted hospital-managed laptops with regulated access to ensure confidentiality. Patients were followed from the date of the first TKI prescribed until the final date of TKI prescription or the end of the study in March 2023.

4.4. Data Collection

Three main categories of data collected were patient characteristics, TKI prescriptions, and factors related to TKI persistency gaps. Data were gathered in three stages using a pre-designed electronic data collection form built on LibreOffice 7.4 (Figure S3) and stored on an encrypted, access-regulated hospital laptop and the hospital’s encrypted, password-protected storage system, ensuring patient confidentiality. A pilot study with dispensing and clinical data from 10 eligible patients tested the data collection form and calculation algorithm.

Patient demographics, including age (recorded as a continuous variable) and gender, were collected using the electronic prescribing system (Iqemo) and cross-referenced with clinical notes from the Trust’s electronic medical records (CWP) to ensure accuracy and maintain anonymity.

Data of all dispensed TKIs for RCC patients, including TKI type, dispensing dates, and prescription coverage period, were collected from the first instance until the final instalment or March 2023 from the pharmacy dispensing system (JAC) to calculate persistence gaps and identify factors influencing adherence. TKI switching was identified and summarised as the proportion of patients using specific TKIs as first-, second-, and third-line treatments. Persistency gaps, defined as prescription refill breaks of 7 days or more [24], were calculated by comparing the days between refills and the prescription’s duration (Figure S4), assuming patient consumption of all TKIs and no secondary sources of supply. These calculations were built into the LibreOffice Base 7.4 collection form (Figure S3), which flagged gaps over seven days.

The primary outcome measure of this study is factors associated with TKI persistency gaps, identified from EMRs (CWP) within one month before and after each gap. This period was chosen as it aligns with typical non-urgent hospital contact intervals. Factors were categorised as ADEs, general clinical factors (e.g., surgery), non-clinical factors (e.g., holidays), and others and collected on a factor discovery form. The common side effects of sunitinib, cabozantinib, and tivozanib were included in the ADE list on this form (Figure S5). The study process flowchart is shown in Figure 4.

4.5. Data Analysis

Data cleaning and checking were conducted to address system recording errors, contradictory clinical notes, duplicates from extended gaps, and gaps with no associated factors (Figure S6).

Descriptive statistics were used to summarise patient characteristics, including age and follow-up duration as median and IQR, gender and TKI prescription types as proportions. In addition, frequencies and percentages of patients at 12- and 24-month follow-up, persistency gaps, and factors associated with persistency gaps were calculated.

Kaplan–Meier analysis, a non-parametric approach assuming data is normally distributed, was used to determine the proportion and time to the first persistency gaps [44] showing risk changes over time. Patients were grouped based on whether their first gap was due to an ADE or a non-ADE factor, and another Kaplan–Meier analysis was performed. When multiple events were recorded for a gap, the research team decided upon the primary reason for stratified analysis. Data was analysed using Microsoft Excel 2023 with an add-in function for the Kaplan–Meier analysis.

5. Conclusions

The study confirms a significant link between ADEs and sub-optimal adherence to TKIs in RCC treatment. With their expertise in medication management, pharmacists are well-positioned to enhance adherence by focusing on managing ADEs, optimising supportive medications, and aligning interventions with patients’ needs and values. Pain-related ADEs, alongside diarrhoea, are likely to affect patients’ quality of life and adherence to TKI therapy. Future research will employ qualitative methods to explore patients’ perspectives, aiming to generate deeper insights into their experiences and support needs.