Physician risk perceptions and surveillance practices for tyrosine kinase inhibitor long-term effects in pediatric CML

Abstract

Chronic myeloid leukemia (CML) is effectively treated with long-term tyrosine kinase inhibitor (TKI) therapy, yet little is known about risks of prolonged TKI exposure in young patients, and long-term effect monitoring is not standardized. We surveyed North American pediatric oncologists (n=119) to evaluate perceived risk of and surveillance practices for potential toxicities associated with prolonged TKI exposure in children and adolescents/young adults (AYAs) with CML. Survey domains included general and specific risk perceptions and surveillance practices for asymptomatic patients on chronic TKI therapy. We analyzed data descriptively and explored relationships between risk perceptions and surveillance. Risk perceptions varied among oncologists but were similar across six categories (thyroid, cardiac, vascular, metabolic, fertility, psychologic), with less than one-third rating each risk as moderate or high in pediatric and AYA patients. More oncologists perceived moderate or high risk of growth abnormalities in children (62% pediatric, 14% AYA) and financial toxicity in all patients (60% pediatric, 64% AYA). A greater proportion of oncologists with moderate or high perceived risk of thyroid abnormalities reported testing thyroid function compared to those with lower perceived risk; patterns for metabolic risk/lipid tests and cardiac risk/tests were similar. In summary, we found that pediatric oncologists had variable risk perceptions and surveillance practices for potential toxicities associated with prolonged TKI exposure. Standardizing surveillance would help quantify risks and refine recommendations.

Introduction

Chronic myeloid leukemia (CML) is diagnosed in 1–3 per million children and adolescents and 5–7 per million young adults (ages 20–29 years) annually in the United States¹. While CML accounts for less than 10% of pediatric leukemias², the prevalence is increasing due to high survival rates with modern targeted treatment. BCR-ABL tyrosine kinase inhibitors (TKIs), which target the oncogene driving CML, have revolutionized treatment. Most patients are treated with single-agent oral TKIs (e.g., imatinib or dasatinib) and do not require intensive chemotherapy or hematopoietic stem cell transplantation. Although treatment-free remission is attainable for about 50% of adults with CML who have an optimal response to TKI, stopping is not recommended for children and adolescents outside clinical trials^3,4. Thus, children and adolescents with CML are exposed to TKI therapy chronically and require years, even decades, of treatment.

TKIs are generally well tolerated in terms of side effects reported in pediatric clinical trials^5–8, yet the risks of prolonged exposure in young patients are not well established. Furthermore, whether and how often to monitor for potential toxicities associated with long-term continuous TKI exposure (long-term effects) is not standardized. The National Comprehensive Cancer Network CML guidelines recommend monitoring growth closely but discuss no other surveillance of long-term effects in children with prolonged TKI exposure³. While experts have published surveillance recommendations within CML review papers, there is no consensus about optimal testing modality or frequency^4,9–12. For example, the British Society for Haematology recommends “regular” thyroid testing, bone density exams every 5 years, and yearly echocardiograms for children on TKIs for CML¹³. This is similar to a 2014 International BFM Group CML committee publication¹² and partially overlaps with a 2019 Children’s Oncology Group (COG) CML working group publication¹⁰, while another recent publication recommends “regular” thyroid surveillance but no other asymptomatic testing⁴.

With incomplete data and inconsistent expert recommendations, little is known about how physicians who prescribe TKIs for young patients perceive the risk of long-term health effects or monitor for toxicities over time. To address these knowledge gaps, we evaluated pediatric oncologists’ perceived risk of and surveillance practices for potential long-term effects of TKIs in children and adolescents/young adults (AYAs) with CML.

Materials and Methods

Study methods

An anonymous cross-sectional Internet-based survey was distributed by email to all COG site principal investigators (PIs) (n=223), who were asked to forward the email to colleagues who treat patients with CML. This snowball sampling approach was selected to maximize sample size in the context of a rare target population (pediatric oncologists who treat CML), as it was expected that some COG site PIs would not have experience treating CML¹⁴. The first two survey questions screened respondents to ensure they were pediatric oncologists who treat CML with TKIs (Supplemental File). The COG communications committee sent initial survey invitations to COG site PIs in November 2019, followed by five email reminders over 3 months through January 2020. This study was reviewed and designated exempt by the Stanford University Institutional Review Board (protocol # 53001).

Surveys were administered using REDCap electronic data capture tools hosted at Stanford^15,16. REDCap is a secure, web-based software platform designed to support data capture for research studies. The Stanford REDCap platform (http://redcap.stanford.edu) is developed and operated by Stanford Medicine Research IT team.

Survey domains

The survey was developed by pediatric oncologists in the COG CML Working Group. As there were no previously validated instruments to measure the outcomes of interest, survey questions were investigator-developed and reviewed by colleagues within the CML Working Group for content and clarity.

The following domains were included in the 35-item survey (Supplemental File):

Respondent characteristics:

Respondents were asked to report their CML experience, measured by number of patients with CML (<5, 5–−10, 11–20, >20) personally treated with TKI therapy during one’s career; general oncology experience, measured by time in practice (trainee/fellow, <10 years, 10–20 years, >20 years); and practice setting (academic, group practice, private practice, other). These three domains were adapted from another pediatric CML physician survey¹⁷. Respondents were also asked which TKIs they had prescribed to patients with CML.

Risk perceptions:

Respondents were first asked to rate the overall risk of TKI-associated long-term effects in pediatric or AYA patients taking any TKI for ≥5 years based on their clinical experience. For the purposes of this study, AYA referred to patients ages 15–29 years. The next questions pertained to specific long-term effects: growth abnormalities, thyroid abnormalities, cardiac disease, vascular disease, metabolic abnormalities, fertility problems, psychologic/emotional problems, financial toxicity, and other (free text write-in). Overall and specific risks were rated separately for children (pre-pubertal) and AYAs (post-pubertal) using the same scale (no risk, low risk, moderate risk, high risk, insufficient data to estimate).

Surveillance practices:

Respondents were asked to select which surveillance studies they routinely monitor (meaning, evaluate more than once over time) in patients on long-term TKI, and whether they do so for each TKI (e.g., imatinib, dasatinib, nilotinib, bosutinib, ponatinib). They were asked about 9 specific surveillance tests: longitudinal assessment of linear growth, pubertal status (Tanner staging), thyroid function, electrocardiogram (EKG), echocardiogram, lipid panel, bone health labs (calcium, phosphorous, vitamin D), bone density, and other (free text write-in). Those who never screen asymptomatic patients were asked to select “don’t screen.”

Clinical practice patterns:

Respondents were asked whether the risk of long-term effects of TKI influences their choice of TKI for newly diagnosed patients with chronic-phase CML, whether they refer patients with chronic-phase CML in major molecular remission to a survivorship clinic while still taking TKI, and whether they transition young adult patients to an adult hematologist/oncologist, as well as their criteria for referral/transition if applicable.

Statistical analysis

Descriptive statistics including frequency and percentages were computed for categorial variables. Bar graphs were used to depict oncologists’ perceived risk of TKI-associated long-term effects in pediatric and AYA patients, and the number of surveillance tests they order in asymptomatic patients. Risk perceptions and surveillance practices were graphed according to CML experience; for these charts, 11–20 patients and >20 patients were collapsed into one category due to low numbers in the >20 patient category. Relationships between risk perceptions and surveillance practices were explored for 3 specific pairings: thyroid abnormalities–thyroid function tests, metabolic abnormalities–lipid panel, cardiac disease–EKG/echocardiogram. For graphs pertaining to surveillance studies, testing practices were shown separately for imatinib and dasatinib (the most commonly prescribed TKIs) and limited to oncologists who reported prescribing a given TKI. The intent of this study was to describe perceived risk of and surveillance practices for potential long-term effects of TKIs; therefore, no formal research hypothesis testing was conducted. Statistical analyses were performed with SAS version 9.4 (SAS Institute: Cary, NC, USA) and graphs were designed in Microsoft Excel and R version 4.0.3 (R Foundation for Statistical Computing: Vienna, Austria).

Results

Characteristics of respondents

A total of 123 pediatric oncologists responded to the survey. Four did not treat CML patients (survey terminated after first question), leaving 119 for this analysis: 100 (84%) with complete data and 19 who did not answer some questions. CML experience varied, with 32% treating <5 patients with TKI, 38% treating 5–10, and 14% treating >10 patients over their career (Table 1). Most had been in practice ≥10 years and worked in academic settings (Table 1). Nearly all respondents had prescribed imatinib (109/119, 92%) and dasatinib (106/119, 89%), while one-third had prescribed nilotinib (39/119) and a minority had prescribed ponatinib (19/119, 16%) or bosutinib (8/119, 7%) (not mutually exclusive).

Table 1.

Clinical characteristics of survey respondents

	All (N=119)
	N	(%)*
Number of patients with CML
personally treated with TKI
< 5	38	(32)
5–10	45	(38)
11–20	12	(10)
> 20	5	(4)
Did not answer	19	(16)
Years in practice
Trainee/Fellow	0	(0)
< 10 years	24	(20)
10–20 years	43	(36)
> 20 years	32	(27)
Did not answer	20	(17)
Practice setting
Academic hospital/clinic	83	(70)
Group practice/clinic	13	(11)
Private practice	3	(3)
Other (write in: urban hospital)	1	(1)
Did not answer	19	(16)

Abbreviations: CML = chronic myeloid leukemia; TKI = tyrosine kinase inhibitor

^*May sum to >100% due to rounding

Perceived risk of TKI-associated long-term effects

Most pediatric oncologists perceived the overall risk of TKI-associated long-term effects as low or moderate in pediatric (78%) and AYA (88%) patients taking any TKI for ≥5 years. A minority reported having insufficient data to estimate risk (18% pediatric, 11% AYA) or perceived the risk as high (2.5% pediatric, 0% AYA) or nonexistent (0.8% pediatric, 0.8% AYA). Overall risk perceptions varied across levels of CML experience (Fig. 1). The proportion of oncologists perceiving a low risk for pediatric patients taking TKIs was lower in those with more CML experience than those with less experience (Fig. 1A). A similar trend was seen for AYA patients (Fig. 1B).

Figure 1. Perceived risk of any long-term effects according to pediatric oncologists’ CML experience, for pediatric and AYA patients taking any TKI for ≥5 years.

The vertical axis represents the number of oncologists. The horizontal axis represents the oncologists’ CML experience defined as the number of CML patients personally treated with TKI therapy. The proportion of oncologists perceiving each risk level is depicted for oncologists according to level of CML experience (column percentages) and shown separately for (A) pediatric and (B) AYA patients.

Considering specific potential health effects associated with prolonged TKI exposure, risk perceptions varied among pediatric oncologists but were similarly distributed across thyroid, cardiac, vascular, metabolic, fertility, and psychologic long-term effects (Fig. 2). About one-half of oncologists perceived low risk of each of these effects, fewer than one-third perceived moderate or high risk, and 11–30% thought there was insufficient data to estimate; risk perceptions were similar for pediatric and AYA patients. In contrast, risk perceptions for growth abnormalities varied by patient type, with 62% rating the risk as moderate or high for pediatric patients versus 14% for AYA patients (Fig. 2). Oncologists perceived a moderate or high risk of financial toxicity in all patients (60% pediatric, 64% AYA) (Fig. 2).

Figure 2. Pediatric oncologists’ perceived risk of specific long-term effects in pediatric and AYA patients with CML.

The vertical axis represents the percentage of oncologists perceiving each risk level. The horizontal axis represents risk perceptions. Bar charts depict the proportion of oncologists perceiving each risk level for pediatric patients (green) and AYA patients (blue).

Surveillance practices for specific TKI-associated long-term effects

Nearly all oncologists reported assessing linear growth for patients on chronic TKI therapy, and most reported following pubertal status with Tanner staging (Table 2). Between 51–65% of oncologists reported ordering routine thyroid function tests, cardiac tests (EKG, echocardiogram), and lipid panels for asymptomatic patients on imatinib or dasatinib (Table 2). Oncologists reported ordering bone health labs more frequently (only 21% do not screen) and bone density exams less frequently (57% do not screen) (Table 2). The total number of surveillance tests that each oncologist reported ordering for patients on imatinib or dasatinib varied from 0–9. A higher proportion of those with more CML experience (>10 patients) ordered ≥5 tests compared to those with less CML experience (≤10 patients): imatinib, 81% vs. 61%; dasatinib, 82% vs. 58% (Fig. 3A). The number of surveillance tests also varied with overall risk perceptions: oncologists who perceived moderate or high risk of TKI-associated long-term effects in children tended to order more tests than those reporting lower perceived risk or insufficient data to estimate risk (Fig. 3B); a similar trend was seen for AYAs (Fig. 3C). With respect to three potential TKI-associated long-term effects with corresponding surveillance tests (thyroid abnormalities, metabolic abnormalities, cardiac disease), perceived risk and surveillance practices tracked together for both pediatric and AYA patients. A greater proportion of oncologists who perceived moderate or high risk of thyroid abnormalities reported ordering asymptomatic thyroid function tests compared to those who perceived low or no risk; a similar pattern held for risk of metabolic abnormalities and lipid panel tests, and for risk of cardiac disease and EKGs or echocardiograms (Fig. 4). Among oncologists who thought there was insufficient data to estimate risk, 25–47% still ordered routine thyroid, lipid, and cardiac surveillance tests.

Table 2.

Pediatric oncologists’ surveillance practices for specific long-term effects of TKI in pediatric and AYA patients with CML

	Don’t screen*	Screen on Imatinib†	Screen on Dasatinib‡
Surveillance tests	N (%)	N (%)	N (%)
Longitudinal assessment of linear growth	2 (2)	91 (98)	86 (98)
Pubertal status (Tanner staging)	14 (14)	80 (86)	77 (88)
Thyroid function tests	34 (34)	60 (65)	54 (61)
Electrocardiogram (EKG)	32 (32)	57 (61)	55 (63)
Echocardiogram	39 (39)	51 (55)	49 (56)
Lipid panel	44 (44)	47 (51)	46 (52)
Bone health labs (Ca, Phos, Vit D)	21 (21)	71 (76)	63 (72)
Bone density exam	57 (57)	38 (41)	31 (35)

^*Among N=100 who answered surveillance practice questions

^†Among N=93 who have prescribed imatinib and answered surveillance practice questions

^‡Among N=88 who have prescribed dasatinib and answered surveillance practice questions

Figure 3. Number of surveillance tests ordered according to oncologists’ CML experience and according to overall perceived risk of any long-term effects in pediatric and AYA patients.

The vertical axis represents the number of oncologists. (A) The horizontal axis represents the oncologists’ CML experience defined as the number of CML patients personally treated with TKI therapy. The proportion of oncologists ordering each number of tests is depicted according to level of CML experience (column percentages) and shown separately for patients taking dasatinib and imatinib. (B, C) The horizontal axis represents risk perceptions. The proportion of oncologists ordering each number of tests is depicted according to overall perceived risk (column percentages) and shown separately for patients taking dasatinib and imatinib, and for (B) pediatric and (C) AYA patients.

Figure 4. Pediatric oncologists’ surveillance test practices according to perceived risk of specific long-term effects in pediatric and AYA patients.

The vertical axis represents the number of oncologists. The horizontal axis represents risk perceptions. The proportion of oncologists ordering each surveillance test is depicted according to perceived risk (column percentages) and shown separately for patients taking dasatinib and imatinib. (A) Thyroid function test practices according to risk of thyroid abnormalities in pediatric and AYA patients. (B) Lipid panel test practices according to risk of metabolic abnormalities in pediatric and AYA patients. (C) Electrocardiogram test practices according to risk of cardiac abnormalities in pediatric and AYA patients. (D) Echocardiogram test practices according to risk of cardiac abnormalities in pediatric and AYA patients.

Clinical practice patterns

Twenty-two of 100 pediatric oncologists who answered questions about their clinical practice reported that the risk of TKI-associated long-term effects influenced their choice of TKI in newly diagnosed patients. Fifteen reported referring patients in major molecular remission to a survivorship clinic while still taking TKI, while 74 do not refer and 11 have no such clinic available. About three-quarters of pediatric oncologists (76/99) transition their young adult patients to an adult hematologist/oncologist. The reported age for transition ranged from 18–30 years, with several oncologists mentioning milestone-based transitions (i.e., after college) and many noting there is variability based on patient factors.

Discussion

In this survey of 119 pediatric oncologists, we found that perceived risk of TKI-associated long-term effects in pediatric and AYA patients with CML varies considerably, as do oncologists’ surveillance practices for potential long-term effects in asymptomatic patients. Risk perceptions tracked with surveillance practices, with those who perceived higher risk of long-term effects ordering more tests. The variability in risk perceptions reflects the dearth of studies evaluating long-term risks of TKIs in young patients and the resultant uncertainty about risk among clinicians. Echoing this, only 22% of oncologists in our study reported that the risk of long-term effects of TKIs influenced their choice of TKIs in newly diagnosed patients. The lack of data about long-term risks has also limited the development of evidence-based clinical practice guidelines for surveillance, likely contributing to the variability reported in our study.

Considering oncologists’ perceived risk of specific long-term effects, several findings are notable. Oncologists perceived a greater risk of growth abnormalities in children than AYAs, which is supported by the accumulating literature linking chronic TKI exposure in pre-pubertal patients with delayed linear growth^18–25. For other potential long-term effects, the most frequently perceived risk was “low risk” with a notable minority reporting “insufficient data to estimate.” Interestingly, we observed a tendency for oncologists with more CML experience to perceive greater overall risk for pediatric patients than those with less CML experience. While financial toxicity is not strictly a long-term health effect, it refers to any problems a patient experiences related to the cost of medical care and is increasingly recognized as an important complication of cancer therapy. Financial toxicity is particularly relevant for new therapeutics such as TKIs, especially when treatment requires daily, indefinite use, as is the case for pediatric CML. In our study, 60% of oncologists perceived a moderate or high risk of financial toxicity for patients on chronic TKIs for CML. Indeed, studies of adults with CML have shown that many patients have high out-of-pocket costs for TKIs, and those with greater financial burden are more likely to stop taking TKIs, which may compromise leukemia outcomes^26,27. Issues of insurance access and coverage are paramount to AYAs with CML who age out of their parents’ insurance while still taking TKIs.

Despite the aforementioned inconsistent surveillance recommendations for young patients on chronic TKIs^3,4,10–13, more than half the oncologists in our study reported ordering routine thyroid, cardiac, and metabolic (lipid panel) surveillance and more than 70% reported monitoring bone health labs in asymptomatic patients with CML. When queried about specific surveillance practices, some oncologists reported ordering many tests in their patients, while others reported ordering far fewer. Accordingly, children and AYAs with CML who are cared for by different oncologists are likely to have variable experiences in terms of whether and how often they undergo asymptomatic testing for potential TKI-associated toxicities. Inter-physician variability may be compounded by patient and health system factors including insurance, access to care, and resource availability. COG CML clinical trials have had variable cardiac and endocrine surveillance requirements for patients treated on study, ranging from none or limited (phase I and II imatinib, phase I dasatinib) to yearly EKG/TSH/DXA (phase II dasatinib, phase II nilotinib); however, the more frequent surveillance on the phase II dasatinib and nilotinib studies reflected secondary aims to evaluate incidence of cardiac/endocrine TKI-associated effects, rather than a recommendation for all patients with CML to receive this intensive monitoring.

We explored relationships between oncologists’ risk perceptions and surveillance practices using three potential long-term effects (thyroid, metabolic, cardiac). In each case, a higher proportion of oncologists who perceived moderate or high risk reported ordering routine surveillance tests compared to those who perceived low or no risk. Association does not imply causation; thus, we cannot determine whether oncologists’ risk perception influenced their surveillance practices or whether ordering surveillance tests (and results of those tests) influenced their risk perceptions, or if both were influenced by another factor. While one would expect surveillance practices to depend upon recognition of organ-specific risks associated with a given treatment, this primarily applies when there is evidence to link a treatment exposure with a specific long-term/late effect risk. In the case of TKIs for CML, emerging data have linked exposure with impaired linear growth; however, other potential TKI-associated toxicities lack a strong evidence base, highlighting the importance of considering both current practices and perceptions of risk among pediatric oncologists prescribing these medications.

CML treatment with chronic, daily, oral TKI therapy represents a new paradigm in pediatric oncology. This contrasts with the more traditional model of cancer care, in which those who are treated with curative intent have a distinct on-treatment phase with chemotherapy and/or radiation therapy and a subsequent off-treatment phase where the focus shifts to monitoring for late effects from prior cancer therapy. Many pediatric oncology programs have survivorship clinics to provide this care. Interestingly, we found that 15% of oncologists in our study refer patients with CML who are in major molecular remission on stable, chronic doses of TKI to a survivorship clinic. While this comprises a minority of the respondents, it highlights the potential role for shared care between oncologists and survivorship experts for patients on chronic cancer therapy²⁸. Patients with CML on long-term TKI therapy may benefit from a comprehensive survivorship evaluation at some point during their treatment (e.g., after 2 years of continuous treatment). While the COG Long-Term Follow-Up guidelines for childhood cancer survivors do not yet include surveillance recommendations for children with chronic TKI exposure^29,30, efforts are ongoing to include targeted therapies in future guideline revisions. In our study, three-quarters of pediatric oncologists reported transitioning patients with CML to be seen by an adult hematologist/oncologist during their young adult years. This underscores the importance of collaboration to ensure we are capturing emerging TKI-associated long-term effects in adults who started taking TKIs in childhood, even if they are no longer cared for by a pediatric oncologist.

We recruited respondents through the COG, the largest pediatric oncology collaborative group in North America, to reach a broad sample of oncologists treating children with CML. While we could not calculate response rate due to the snowball sampling approach, it is encouraging that the characteristics of our respondents are similar to those from a prior pediatric CML survey despite using different recruitment methods¹⁷. Our survey had some missing data because all questions were optional to decrease participant response burden. However, the respondents with missing data (n=19) had a similar distribution of overall risk perceptions and TKIs prescribed as the respondents with complete data (n=100). The limitation of any self-report survey is that reported surveillance practices may not reflect actual clinical practice. That said, a recent retrospective study of cardiac and endocrine screening for potential TKI-associated long-term effects in pediatric CML showed substantial patient-to-patient variability in actual screening practices at two institutions³¹, mirroring the variability in reported surveillance practices in this study.

In summary, the lack of uniformity in pediatric oncologists’ risk perceptions and surveillance practices for potential TKI-associated long-term effects reflects uncertainty about long-term risks and lack of standardized surveillance guidelines. Given the rarity of pediatric CML, innovative data collection methods are needed to identify emerging long-term effects related to TKI therapy. One such approach — the ACCELERATE international prospective long-term follow-up registry for children treated with novel anticancer therapies — is being developed and will provide invaluable insights³². In the near term, it is important to consider how we may standardize clinical practice even in the absence of strong evidence to guide surveillance recommendations, with the goals of collecting data to refine future guidelines and giving patients more similar experiences across diverse practice settings. With increasing use of chronic targeted therapies in CML, there is an opportunity to consider shared oncology and survivorship care models. Because many patients transition from pediatric to adult hematology/oncology care for ongoing TKI treatment, collaborations between pediatric and adult specialists will be essential to detect emerging long-term effects related to these agents.

Data sharing:

The data that support the findings of this study are available from the corresponding author upon reasonable request.