Adverse Drug Reactions Involving Protein Kinase Inhibitors: A French Pharmacovigilance Database Study Comparing Safety in Younger and Older Patients (≥ 75 years) with Cancer

Background Despite the increasing incidence of cancers in elderly people, this population remains under-represented in clinical trials. As a result, data describing the safety and efficacy of protein kinase inhibitors (PKIs) specifically in older patients with cancer are lacking. Advanced age may have a significant impact on drug pharmacology, but data on PKI safety in older patients remain poorly described in “real life”.Objectives We performed an observational study describing adverse drug reactions (ADRs) related to PKIs and compared the results for younger and older (aged ≥ 75 years) patients.Methods We extracted all notifications considered to be related to PKIs from our regional pharmacovigilance database between March 2003 and November 2015. Information about patients, drug exposure, and ADRs were captured. After a descriptive analysis of ADRs, we compared their characteristics between older and younger patients.Results In total, 214 patients experienced 320 ADRs related to PKIs. Slightly over half the patients were male (57%). The mean age was 64.1 years (range 15–90), and 52 (24.3%) patients were aged ≥ 75 years. Cutaneous reactions were the most frequently reported ADRs (22.9%), followed by gastrointestinal (16.8%) and respiratory (12.1%) ADRs. The most often involved PKIs were imatinib [54 patients (25.2%)], dasatinib [25 (11.7%)], sunitinib [25 (11.7%)], erlotinib [25 (11.7%)],and sorafenib [24 (11.2%)], followed by vemurafenib [17 (7.9%)] and everolimus and nilotinib, with 11 patients (5.1%) each. These drugs were administered mostly for three therapeutic indications: chronic myeloid leukemia [56 patients (26.2%)], malignant urinary tract cancer [31 (14.5%)], and bronchial cancer [27 (12.6%)]. Comparison of PKI-related ADRs between older and younger patients showed no significant difference, except for the mean number of coadministered drugs (5.3 vs. 4.2; p = 0.03) and the proportion of vascular ADRs (17.3 vs. 4.3%; p = 0.005), mainly arterial hypertension, which were both significantly more frequent in the older group.Conclusions Our work showed that PKI-related ADRs in older patients were similar to those in younger adults, except for vascular ADRs, which were significantly more frequent in the older population. Our results require confirmation by larger studies but could lead physicians to be more vigilant about cardiovascular comorbidities during initiation of PKIs in elderly people.

The proportion of older patients (aged ≥ 65 years) in the population is increasing. Currently, this age group repre- sents about 18.8% and 8.7% of the French and worldwide populations, respectively [1]. According to France’s national statistical institute, elderly people will represent one-third of the population by 2060 [1]. In Europe and the USA, 69% and 75% of cancer-related deaths, respectively, occur in peo- ple aged ≥ 65 years [2]. These results are relatively similar to those in other western countries. In the next decade, the proportion of new cases in elderly people will continue to increase because of the increasing life expectancy and will represent 70% of the total of new cancers diagnosed. Thus, a geriatric oncology specialization has arisen in recent dec- ades. This initiative consists of a multidisciplinary approach with a specific geriatric assessment and helps oncologists to improve drug safety and compliance [3].
As chemotherapy causes severe adverse drug reac- tions (ADRs) that dramatically impact the quality of life of patients (especially older patients), targeted therapies such as protein kinase inhibitors (PKIs) have become a new modality to treat frail people. However, physicians may be reluctant to treat older patients with PKIs because of the lack of knowledge about their safety. Indeed, older people are rarely included in clinical trials [4]. For example, a recent publication showed that older/frail patients are underrep- resented in randomized controlled trials (RCTs) studying targeted therapies in metastatic colorectal cancer (mCRC) and that elderly patients included in RCTs do not reflect well the general elderly population with mCRC because of the exclusion criteria [5]. As a result, very little data are avail- able in the literature that describe the efficacy and safety of PKIs in older patients with cancer.Therefore, we aimed to conduct a retrospective observa- tional study to define the ADR profile of PKIs notified by health professionals to the Toulouse regional Pharmacovigi- lance Centre (southwest of France). We then compared these ADRs between older (aged ≥ 75 years) and younger patients.

2.1.Data Source
The French pharmacovigilance system is based on a net- work of 31 regional centers of pharmacovigilance (RCPV) that receive all spontaneous ADR reports. Anyone, includ- ing health professionals and patients, may declare an ADR to an RCPV. Since 1985, these reports are registered in the French pharmacovigilance database (FPVD) that now includes > 600,000 ADR reports [6].
For each report, information about the patient (age, sex, past medical history), ADR [type, coded according to the Medical Dictionary for Drug Regulatory Activities (Med- DRA) [7], date of onset, duration and outcome], and drug exposure (name, intake dates, doses) is recorded in the FPVD.A detailed summary of the clinical description is added at the end of each ADR report. For all reports, a clinical and pharmacological analysis is performed and a causality assessment (“imputability” or “imputation” score) is per- formed for each drug using the validated French causality assessment method. If causality is found between the drug and the occurrence of the ADR, drugs are defined as “sus- pected”; if not, drugs are defined as “associated” (non-sus- pected) [8]. A “serious” ADR is defined as “any untoward medical occurrence that at any dose results in death, requires hospital admission or prolongation of existing hospital stay, results in persistent or significant disability/incapacity or is life threatening” [9].

2.2.Data Analysis
In this study, we extracted ADR reports from March 2003 to November 2015 that involved PKIs (i.e., axitinib, bosutinib, crizotinib, dasatinib, erlotinib, everolimus, gefitinib, ibruti- nib, imatinib, nilotinib, ponatinib, regorafenib, ruxolitinib, sorafenib, sunitinib, temsirolimus, and vemurafenib) and were defined as “suspected”.For each ADR, the following information was recorded: the date of notification; sex, age, medical history, and comorbidities of the patient; drug exposure (“suspected” and “associated drugs”, label or off-label use); characteris- tics of ADRs (seriousness, type classified by system organ class according to MedDRA classification, time of onset, outcome, imputability score). We also carefully analyzed the clinical summary of each case to retrieve not-coded informa- tion such as blood creatinine level, clearance of creatinine (if available), and a detailed description of the ADR.

2.3.Statistical Analysis
First, a descriptive analysis was performed: continuous variables are expressed as means ± standard deviations (SDs), and categorical variables are expressed as frequencies and percentages. Then, a comparative analysis between the char- acteristics of older (aged ≥ 75 years) and younger patients was conducted. Categorical variables were compared using the Chi-squared test (if expected frequencies were < 5, Fish- er’s exact test was used). Quantitative data were compared using the Student’s t test or the Wilcoxon test, as appropriate. The significance threshold was 5% (two-tailed). Statistical analyses were performed with SAS® software, version 9.4 (SAS Institute). 3.Results 3.1.General Characteristics of Adverse Drug Reaction (ADR) Notifications From March 2003 to November 2015, a total of 214 ADR notifications involving one or more PKI were reported to the RCPV of Toulouse. Just over half of the patients were male (57%). The mean age was 64.1 ± 13.5 (range 15–90) years. A total of 320 ADRs were described in these 214 notifications (i.e., a mean value of 1.5 ADRs per patient). Table 1 shows the characteristics of the 214 reports. The more frequent PKI-related ADRs were skin and tis- sue disorders [49 reports (22.9%); mainly various types of rash, acne, and hand–foot syndrome], followed by gastroin- testinal disorders [36 (16.8%), mainly diarrhea, vomiting, and pancreatitis] and respiratory and mediastinal disorders [26 (12.1%), mainly pleural effusion and lung infection]. The mean time of onset of PKI-related ADRs was 113.7 ± 243 (range 0–1461) days in the whole population. The number of cases was too small to analyze times of onset according to therapeutic indications.Of the 214 notifications, 162 (75.7%) involved seri- ous ADRs, including 124 (76.5%) hospital admissions or prolongations of existing hospitalization, and 18 (11.1%) were reports of life-threatening ADRs. In total, 14 (6.5%) cases were fatal. Only one ADR (0.5%) corresponded to a drug–drug interaction (DDI) (acute renal failure after dasat- inib and ganciclovir co-administration). In our study, the 214 patients experiencing an ADR were treated with 949 drugs. Among these, 319 were con- sidered “suspected” (220 PKI and 99 other medications, mainly anti-infectious, other chemotherapy, antalgics). The most frequently involved PKIs were imatinib [54 patients (25.2%)], dasatinib [25 (11.7%)], sunitinib [25 (11.7%)],erlotinib [25 (11.7%)], and sorafenib [24 (11.2%)], followed by vemurafenib [17 (7.9%)] and everolimus and nilotinib with 11 cases (5.1%) each. All other PKIs were involved in five or fewer notifications (< 2.3% each). These drugs were administered mostly for three therapeutic indications: chronic myeloid leukemia [56 patients (26.2%)], malignant urinary tract cancer [31 (14.5%)], and bronchial cancer [27 (12.6%)]. 3.2.Results of the Comparative Study As shown in Table 1, among the 214 reports of ADRs, 52 (24.3%) concerned patients aged ≥ 75 years. Sex distribution was the same in the two groups (p = 0.24).Characteristics of ADRs were not significantly differ- ent between the two age groups (blood creatinine level, clearance of creatinine, PKI used, therapeutic indication, time of onset, comorbidities, particularly vascular comor- bidities) except for two characteristics: the older group (age≥ 75 years) used significantly more drugs than the younger group (5.3 ± 4.0 (range 1–18) vs. 4.2 ± 3.8 (range 1–28); p = 0.03). Significantly more PKI-related vascular ADRs were reported in the older group (17.3 vs. 4.3%; p = 0.005), mainly arterial hypertension, occlusive disorders, and bleeding.Although the outcomes and seriousness of ADR notifica- tions were comparable, 13 patients from the younger group and one in the older group died. This difference was not significant (p = 0.20). 4.Discussion To the best of our knowledge, our study is the first to use data notified by health professionals to investigate PKI- related ADRs in real-life practice and to compare these ADRs between two populations, those aged < 75 and those aged ≥ 75 years. Although “elderly” is often defined as peo- ple aged ≥ 65 years (World Health Organization definition), we used an age limit of 75 years, which represents a frailer population in real life. Despite the increasing incidence of cancers in older people, older people (especially those aged ≥ 75 years) are often under-represented in clinical trials. This is particularly true in clinical trials assessing targeted therapies, with one analysis of 19 such studies revealing that only 32% of all patients were aged ≥ 65 years [10]. No sig- nificant safety data are available for patients aged > 85 years [11]. Overall, our study shows, in routine clinical practice, few qualitative and quantitative differences in PKI-related ADR characteristics between younger and older patients with cancer.The lack of significant results may have various explana- tions. First, the sample size was small, and subgroup analysis of a sample of 214 cases has low statistical power. Second, the source of data was a pharmacovigilance database, the main limitation of which is under-reporting. It is not possible to know the true incidence of ADRs because the total level of exposition remains unknown [12]. However, since our two age groups were sampled from the same database, we can hypothesize that the rate of under-reporting was the same in both groups. Moreover, it is generally accepted that under- reporting should be reasonably similar for similar drugs sharing the same indication, country, and marketing period[13].

Finally, the third limitation is that retrospective studies carry the possibility of missing data. Important parameters, such as performance scores (i.e., Charlson comorbidity score or activities of daily living score), ADR severity (or grade), and information on drug doses and blood creatinine levels, were not sufficiently available in the spontaneous reports because they are not systematically recorded in the FPVD. Despite these limitations, compared with previous publi- cations about the safety of PKIs in older patients (mainly subgroup analyses of clinical trial results), our study using regional French ADR reporting provides safety results that are more in line with real medical practice and therefore constitutes another approach to improve our knowledge on the circumstances surrounding the occurrence of these little- known ADRs.It is difficult to compare our results with existing litera- ture since most available safety data come from subgroup analyses of clinical trials and not from post-marketing data. Given the wide range of PKIs used in the elderly, the dif- ferent classes of PKIs, the different indications for use with very different natural courses of various cancers, and the heterogeneity of the elderly definition (according to age or to geriatric assessment), comparing safety data and interpreting and translating research findings on the impact of PKIs in older patients remains difficult. As older people included in clinical trials are not representative of the entire cohort of older people with cancer, the results cannot be generalized for all geriatric patients with cancer [10].

Few studies assessed the global safety of PKIs in older patients with cancer. When available, these studies often focused on the efficacy and safety of one PKI (or one type of PKI) in one type of cancer and reported that, glob- ally, a slight increase in ADRs is sometimes observed in older patients, whereas efficacy is mainly comparable with younger patients. However, all authors highlighted the need for close monitoring of older patients to prevent and monitor potential PKI-related ADRs [14, 15]. An adapted geriatric selection could help decrease ADRs, and specific clinical trials for older patients with cancer would be useful [10].In our sample, among the 214 ADR reports, 52 (24.3%) concerned patients aged ≥ 75 years, with no difference in terms of sex. In addition to under-reporting, this small sam- ple may be explained by reluctance on the part of physicians to prescribe PKIs to patients aged ≥75 years due to a lack of safety data in this age group.As described in the literature [16], we found that older people used significantly more medicines than younger peo- ple (on average, 5.3 vs. 4.2 drugs per patient, respectively). Surprisingly, this greater medicine consumption in the older group did not seem to increase the proportion of ADRs or DDI occurrences, as only one ADR (0.5%) was defined as a DDI in our study. Therefore, this specific point differs from the findings of several other studies. Polypharmacy is known to be an increasing problem in the care of polymor- bid patients (who are often older) and increases morbidity, mortality, hospitalization rates, and drug-related problems [17–19]. For example, Field et al. [16] showed that the risk of ADRs is three times higher in patients taking at least seven medicines. Because of their numerous comorbidities, elderly people are often exposed to a combination of a large number of drugs. This exposes them to DDIs, especially with the PKIs eliminated via cytochrome P450 (CYP450). Many other drugs can be induced or inhibited by CYP450, and these DDIs expose patients to a risk of increased toxicity or decreased PKI efficacy [10].

In addition, despite the polypharmacy found in many older patients, our study reveals that the seriousness of ADRs was the same in the two age groups, even though increasing age is known to be a risk factor for serious ADRs [20].As our results indicate, the mean time of onset of PKI- related ADRs was 113.7 ± 243 (range 0–1461) days in the whole population, with no significant difference between younger and older patients. The number of cases was too small to analyze times of onset according to therapeutic indications. Unfortunately, data about drug discontinuation and/or dose reduction were missing. Patients experiencing PKI-related ADRs represent a heterogeneous group, and the timing of ADR onset could change according to treat- ment duration. However, data on PKI-related ADRs as a function of treatment duration are conflicting. For example, some authors have shown that patients receiving imatinib 400 mg for 36 months experienced more grade 3 ADRs and stopped the drug more frequently than those who received the drug for 12 months. By contrast, other studies showed that some ADRs occurred more frequently at the beginning of treatment and were less severe and frequent over time. Finally, ADRs such as haematological reactions, skin rash, or nausea improved with dose escalation during the course of the treatment [21].

In terms of the type of ADRs, skin and subcutaneous ADRs (22.4%) were the most commonly reported in the two age groups. PKIs of epidermal growth factor receptor (i.e. erlotinib, gefitinib) and PKIs of BRAF (i.e., regorafenib, vemurafenib) were the most often involved in cutaneous ADRs (mainly rashes, acneiform reactions, and hand–foot syndrome). These results are comparable with those described in the literature showing that cutane- ous reactions are frequently reported with PKIs [22, 23]. Gastrointestinal ADRs (such as nausea, vomiting, and diar- rhea) ranked second in both age groups. These ADRs are not specific to PKIs, but mild to moderate nausea, vomiting, diarrhea, constipation, and abdominal pain are frequent with all PKIs. Even if they are known, digestive ADRs can be very serious; our study found two digestive perforations with fatal outcomes were reported in young patients treated with sorafenib. Serious gastrointestinal ADRs were found in third place in the study by Faye et al. [22].In a recent review, Daste et al. [10] assessed the safety and efficacy of targeted therapies in elderly patients with solid tumor cancers, principally from data of pivotal clini- cal trials with subgroup analyses: the safety and efficacy of targeted therapies in elderly patients seems similar to that in younger patients, with an increase in the incidence of specific ADRs in elderly patients with selected PKIs. In comparison with younger patients, the increased ADRs in elderly patients were mainly fatigue, peripheral edema, ano- rexia, anemia, increased liver enzymes, and rash. Globally, general and digestive ADRs were more frequently reported in older patients [10].

Other studies have also shown that the frequency of PKI- related ADRs was no greater than in younger patients but highlighted the potentially greater severity in older patients, with a possible decreased quality of life and a risk of “dom- ino pathologies” [12, 15, 24].In our study, we showed that older patients with cancer, compared with younger patients, had significantly more vas- cular ADRs. The significantly higher rate of vascular reac- tions was not explained by higher prevalence of vascular comorbidities in older patients. These ADRs were mainly arterial hypertension, occlusive events, and hemorrhages. These ADRs are well-known complications of angiogene- sis-inhibitor drugs (i.e., sunitinib, sorafenib). For example, hypertension of any grade was reported with an incidence ranging from 17 to 49.6% in RCTs [25], but recent reviews did not indicate an increased incidence of vascular ADRs in the elderly subgroup [10, 15]. Finally, concerning hem- orrhages, no studies reported a higher frequency in older patients. However, it seems fairly clear that older people have an increased risk of bleeding because of the more fre- quent use of antiplatelet and anticoagulant drugs in the pre- vention of thromboembolic events [10].

Since very few data are available in the literature, our results provide insight into the risk of ADRs with PKIs in patients aged ≥ 75 years in real life. First, we showed that PKI-related ADRs in older patients were similar to those in younger adults Then, the most frequent ADRs were the same in the two populations (cutaneous and digestive). Only the incidence of vascular ADRs was significantly higher in older patients. Our results require confirmation by larger observational studies or prospective studies within hospital departments. They could lead physicians to be more vigilant about cardiovascular PLX4032 comorbidities during initiation of PKIs in older patients with cancer.