Effect of the patient information brochure in communicating the risks associated with crizotinib treatment to patients with non‐small cell lung cancer (NSCLC) in Europe

Abstract Crizotinib (XALKORI®) is indicated for anaplastic lymphoma kinase‐positive and ROS1‐positive advanced non‐small cell lung cancer. This study evaluated the distribution of the crizotinib patient information brochure (PIB) in Europe and patient knowledge of the key messages in the PIB. A cross‐sectional survey was conducted in 10 European countries among patients who received crizotinib to ascertain whether each patient received and read the PIB, and his/her knowledge of its key messages on hepatotoxicity, interstitial lung disease/pneumonitis, QTc prolongation, bradycardia, and vision disorders. Of the 341 patients contacted, 40 responded (11.7%), and 39 patients were eligible. A total of 77% of respondents acknowledged receiving the PIB, of which, 93% reported reading it. Knowledge of the individual side effects ranged from 36% to 85%, and precautions for use ranged from 56% to 67%. Understanding the reasons for calling a physician ranged from 54% to 85%. Knowledge of each of the 6 key side effects was greater among readers of the PIB compared to non‐readers or respondents who did not recall receiving the PIB. Approximately three‐quarters of survey respondents recalled receiving the crizotinib PIB and respondents who read the PIB were more knowledgeable of the key side effects of crizotinib than those who did not read or receive. Caution should be taken in generalizing these results because of the potential for selection bias and small sample size. These survey results suggest that the crizotinib PIB may be an effective risk communication tool for crizotinib‐treated patients in Europe.


| INTRODUC TI ON
Lung cancer is the leading cause of cancer-related mortality worldwide. 1 In 2018, the number of new lung cancer cases was estimated at 2.1 million worldwide, representing 11.6% of all new cancers, and the number of lung cancer deaths was 1.8 million, representing 18.4% of the total cancer deaths. 1  Crizotinib has been associated with a number of safety risks including hepatotoxicity, interstitial lung disease /pneumonitis, QTc prolongation, bradycardia, and vision disorders. The crizotinib label lists these risks as adverse reactions, and in Europe, these risks are included in the patient information leaflet (PIL). Additionally, Pfizer has developed educational materials in Europe as part of additional risk minimization measures (RMMs) requested by the Committee for Medicinal Products for Human Use (CHMP), which include a patient information brochure (PIB) to further inform patients receiving crizotinib treatment about known risks associated with crizotinib, as well as a patient identification (ID) card. The patient ID card, which includes spaces for patients to add their name, their oncologist's name, and the date crizotinib was started, is provided for patients to show their other healthcare providers.
This study was designed to assess the effectiveness of both the crizotinib PIB and patient ID card among EU patients. The specific objectives of the study were to assess patients' awareness, receipt, and use of the crizotinib PIB and patient ID card, and to assess if patients' knowledge of the key risks and actions required to minimize the key risks was in accordance with the information provided in these materials.  regulations. Pre-testing was not conducted in Austria since the instrument had been tested in German. Since the European Medicines Agency (EMA) had endorsed the English version of the questionnaire, it was not pre-tested in Ireland or the UK. Pfizer offices in Austria, Ireland, and the UK reviewed the survey questionnaire to confirm that terminology used was consistent with local crizotinib educational materials.

| Survey instrument
Experienced personnel in cognitive pre-testing and linguistic validation of survey questionnaires conducted pre-testing with 1-on-1 interviews. The cognitive pre-test informed necessary minor revisions to most of the country-specific versions of the patient questionnaire, mainly modifications of the initial translations to conform to local standards or language nuances (eg, in some countries, "true/false" was more commonly communicated as "yes/no"). Other changes identified from pre-testing included: • The generic name for XALKORI® (ie, crizotinib) was provided throughout the survey because the brand name was not recognized by several patients • The French translation of instructions regarding side effects associated with XALKORI® was revised to make clear that the purpose was to inquire about information the respondent learned from the PIB rather than to collect information about side effects personally experienced Surveys were self-administered in local languages either via the internet or paper, depending on respondent's preference. Confirmit, a software platform designed specifically for surveys, was used to administer the survey and collect data.

| Survey administration
After securing any required local approvals (such as ethics committees), physicians who were willing to recruit patients for the survey were provided with patient survey kits. The contents of the patient survey kit were: • a letter to patients inviting participation in the survey which included study details, a unique code, and instructions for online access of the survey, • for all countries except France, an informed consent document; in France, a study information document conforming to local regulations in lieu of the informed consent, • a paper survey with the same unique code to be completed by the patient and a postage paid, pre-addressed envelope for returning the paper survey, if the paper survey method was chosen by the patient.
The number of completed surveys was tracked to monitor progress to identify study sites and physicians with no or few surveys completed by patients.

| Data analysis
The dataset for analysis comprised all eligible patients who answered  percentages were determined for the effectiveness endpoints using exact methods. Familiarity with the key crizotinib side effects ranged from 36% to 85% (Table 2). Most respondents expressed knowledge of "changes to vision" (85%), "dizziness, light-headedness, fainting, tiredness" (69%), and "abnormalities in liver blood tests" (61%).

| RE SULTS
Nearly half (49%) of respondents knew crizotinib was associated with "chest discomfort or irregular heartbeat." More than a third of respondents knew that crizotinib may "slow heart rate" (38%) or could cause "breathing problems" (36%).
In general, knowledge of precautions for crizotinib use was higher than knowledge of side effects. About two-thirds of respondents reported knowledge of the necessity of stopping driving and operating machinery for changes in vision (67%) or informing their physician about persistent or worsening visual changes (69%). More than half of respondents (56%) reported knowledge that their heart function would be monitored by their doctor and that the dosage of crizotinib might require adjustment.
The range of understanding of reasons to contact the physician ranged from 54% to 85%. Specifically, the rates of knowledge of when to contact the physician for "light-headedness, chest discomfort, fainting" and "difficulties with breathing, cough, fever" were 85%. Most respondents (74%) reported knowing to contact the physician for "nausea, vomiting," 67% knew to contact the physician for "skin and whites of your eyes turn yellow," 56% knew to contact the physician for "urine turns dark or brown (tea colour)", and 54% knew to contact the physician for "itching, or bruised more easily than usual." Knowledge of individual side effects stratified by respondents who did and did not read or receive the PIB is presented in Figure 1. Respondents who reported reading the PIB were more knowledgeable of each of the 6 key side effects than respondents who did not read or receive it. Awareness of the side effects and precautions for use of crizotinib, and reasons for contacting the physician was high for vision-, hepatic-, and dizziness-related items, with knowledge levels ranging from 61% to 85%. However, fewer than half of respondents knew crizotinib was associated with "chest discomfort or irregular heartbeat," "slow in heart rate," or "breathing problems." Knowledge levels in our study were similar or higher to those reported by Landsberg et al, where knowledge levels of risks associated with aripiprazole ranged from 46% to 69% and knowledge levels of behaviors in case these risks occur was 56%-69%. 6 Similarly, a study that evaluated knowledge levels of precautions for use for, and risks associated with, varenicline reported patient knowledge levels of 19%-82%. 7

| D ISCUSS I ON
Knowledge levels in our study were lower than those reported by Brandenburg et al, where knowledge levels of teratogenic-related risks for lenalidomide and thalidomide assessed by 5 survey questions ranged from 87% to 98%. 8 However, it is not surprising that knowledge levels for teratogenic-related risks were high, given the severity of this risk as well as the restricted distribution REMS program in the US specifically employed to minimize teratogenic risks associated with lenalidomide and thalidomide.
Ideally, the effectiveness of RMM should be measured against a comparator group not exposed to the RMM. However, such a comparator group is not possible for medicines that have RMM required at the time of initial authorization. 9 In order to circumvent this limitation, we considered levels of knowledge for all 6 key side effects by whether patients reported having read or received the PIB or not. In this study, respondents who read the PIB showed consistently greater knowledge for all 6 key side effects than respondents who did not read or receive, suggesting the PIB was a Since ALK-positive NSCLC is relatively rare, the number of physicians prescribing crizotinib is expected to be low. Our study faced recruitment challenges largely because of the paucity of treating physicians who were available to identify patients for survey participation, and feasibility limitations such as the undue lengthy processes for ethics approvals in some countries without prior experience with surveys that evaluate the effectiveness of RMMs.
Hence, based on the confidence interval for one proportion with exact (Clopper-Pearson) formula, the small number of patients who completed the survey (n = 39) resulted in low precision of the knowledge rate estimates (11.9%-15.5%). Similar to recommendations provided by Landsberg et al, 6 we suggest that additional guidance from regulatory agencies is needed to improve the collaboration between the industry and patient stakeholders to facilitate wider reach to assess the clinical importance of patient-directed interventions such as RMMs.
Our study used a convenience sample, which could have introduced a selection bias. We sought to limit this effect by including countries with the highest crizotinib prescribing rates and obtain- An additional source of potential bias is the use of self-reporting. Although self-reporting is the only way to assess a person's knowledge, patient-reported information may still be subject to recall bias.
Overall, the survey results suggest that most patients who responded to the survey received and read the crizotinib PIB. The majority of patients who responded to the survey were aware of crizotinib side effects, precautions for use, and reasons to contact the physician. Knowledge rates were consistently greater among patients who read the PIB compared with patients who did not read or receive the PIB. The findings of this survey suggest that the PIB may be an effective way to provide information about risks to patients receiving crizotinib.