A retrospective observational study was conducted in the period January 2006-December 2021 to evaluate the safety of drugs approved for the treatment of NSCLC in a real-world scenario. In Italy, all suspected ADR reports are collected through the RNF. The RNF was established in 2001 by the Italian Medicines Agency (Agenzia Italiana del Farmaco, AIFA) with the aim of collecting all suspected ADRs from drugs and vaccines reported by health professionals and citizens from all Italian regions. According to Italian pharmacovigilance standards, all Italian regions, including Sicily, have full access to their own regional spontaneous reporting data in the RNF and carry out post-marketing surveillance activities through the Pharmacovigilance Regional Centers (PRCs). Since July 2017, AIFA has created a public online access system called RAM monthly updated which allows to check all aggregated data related to national ADR reports registered since 2002 into the RNF. The research in the RAM system can be performed by the trade name or by the active substance indicated as suspected. Reports can be consulted by year, seriousness, age group, gender, while ADRs by the Medical Dictionary for Regulatory Activities (MedDRA^®) System Organ Class (SOC) and Preferred Term (PT).

In the RNF, PRCs can analyze all detailed regional ADR reports. Each ADR report contains information on the patient, a detailed description of ADR with the relative time to onset (TTO), seriousness, outcome, dechallenge, rechallenge, and relevant laboratory tests, data related to suspected and concomitant drugs with their dosage, frequency, route of administration and therapeutic indication, and all potential comorbidities. Drugs are codified using the Anatomical Therapeutic Chemical (ATC) classification, while suspected ADRs are grouped according to the MedDRA^® (16). Suspected ADRs are classified as serious if they were life-threatening or fatal, required hospitalization, led to persistent or significant disability, or a congenital anomaly/birth defect or have been defined as other medically important conditions (17). The TTO was expressed in days and was calculated as the time elapsed between the start of drug use and the beginning of the ADR.

All national aggregated ADR reports collected into the RAM system from the date of drug approval until December 31, 2021 and related to the following TKIs approved for NSCLC were consulted: AFT, ALEC, BRG, CER, CRIZ, ERL, GEF, LORL, NTB, and OSI. Moreover, only cases related to the branded name “Vargatef^®” for NTB were included to avoid some therapeutic bias with other available medications having NTB as active substance and approved for other indications. Entrectinib and selpercatinib were excluded from the study due to their recent approval for the treatment of NSCLC, while larotrectinib was not included because it is not yet approved in Italy.

Furthermore, all Sicilian detailed ADR reports collected into the RNF database and having as suspected drug one of the above mentioned TKIs and signaled from January 2006 to December 31, 2021 were collected. Before to proceed with the analysis, literature data and duplicates were excluded from the database.

A descriptive analysis of publicly available aggregated national data coming from the RAM system and of regional full access detailed reports was performed to assess demographic characteristics and drug-related variables. Specifically, analyses for year, sex, age, and seriousness were conducted. All ADRs were analyzed according to MedDRA^® classification levels for SOC and PT by grouping each ADR into the corresponding SOC and synonymous PTs of the same clinical condition under a unique term ( Supplementary Table S1 ). Absolute and relative frequencies were calculated for all variables.

Consequently, a case-by-case analysis of all detailed reports registered in Sicily was carried out, considering any concomitant drugs, potential comorbidities, outcomes, TTO and the causality assessment. The causal relationship between the ADR and the suspected drug was estimated using the Naranjo’s algorithm, classifying ADRs in highly probable (score ≥ 9), probable (score 5 - 8), possible (score 1 - 4), or doubtful (score ≤ 0) (18).

This study included a total of 3,048 ADR reports collected from January 1, 2006 to December 31, 2021 into the RAM system and having as suspected drug one of TKIs used for the treatment of NSCLC. A gradual increase of reports was noted over the years with a decrease in 2014 (n = 130; 4.3%) and a peak in 2019 (n = 338; 11.1%) ( Figure 1 ). The trend showed a higher number of reports related to EGFRi: from 2006 to 2016 to ERL (n = 1,352; 44.5%), and from 2017 to 2018 to AFT (n = 151; 5.0%), while in the last 3 years most of reports concerned OSI (n = 240; 7.9%). Focusing on seriousness, the majority of ADRs were not serious (not serious, 71.9% vs. serious, 26.7%) and the highest number of serious ADRs (SADRs) was recorded for NTB (n = 27; 73.0%), OSI (n = 156; 56.5%), and for the ALKi LORL (n = 6; 60.0%) and ALEC (n = 76; 54.3%). Considering patients’ characteristics, ADR reports were slightly more frequent in females (females, 52.2% vs. males, 46.6%) although a higher proportion of males was reported for NTB (n = 24; 64.9%), ERL (n = 893; 61.7%), and LORL (n = 6; 60.0%); the age group >65 years was mainly reported (n = 1,617; 53.0%), but the age group 18-65 years was the most shown for ALKi: ALEC (n = 82; 58.6%), BRG (n = 7; 70.0%), CER (n = 32; 61.5%), CRIZ (n = 168; 61.3%), and LORL (n = 20; 54.1) ( Table 1 ).

Considering ADRs, they were mostly related to skin and subcutaneous tissue disorders (n = 1,766; 57.9%), gastrointestinal disorders (n = 1,024; 33.6%), general disorders and administration site conditions (n = 536; 17.6%), infections and infestations (n = 483; 15.8%), hepatobiliary disorders (n = 419; 13.7%), and respiratory, thoracic and mediastinal disorders (n = 309; 10.1%). Skin and subcutaneous tissue disorders, especially rash, were mostly reported for EGFRi ERL (n = 1,120; 77.3%), AFT (n = 277; 63.7%), and GEF (n = 214; 58.5%). The onset of gastrointestinal disorders, especially diarrhea, were mainly reported for NTB and AFT (n = 31; 83.8% and n = 268; 61.6%, respectively). Furthermore, general disorders and administration site conditions were particularly shown for NTB (n = 20; 54.1%), LORL (n = 4; 40.0%), and ALEC (n = 48; 34.3%). Concerning infections, including folliculitis, the EGFRi ERL and AFT had a higher number of related ADRs (n = 296; 20.4% and n = 79; 18.2%, respectively). The onset of hepatobiliary disorders, mainly hypertransaminasemia, was detected for ALKi CER (n = 31; 59.6%), ALEC (n = 57; 40.7%), and CRIZ (n = 98; 35.8%), but also for NTB (n = 12; 32.4%) and GEF (n = 111; 30.3%). More than half of ADRs related to blood and lymphatic system disorders were associated with the VEGFi NTB (n = 19; 51.4%) ( Table 2 ).

Considering regional data, 263 Sicilian reports related to the use of TKIs for the treatment of NSCLC were collected. On the contrary to Italian reports, the regional ADR reporting rate has not gradually increased over the years: after a considerable peak in 2010 (n = 26; 9.9%) followed by a decrease in 2011 (n = 11; 4.2%), other peaks were noted in 2016 (n = 35; 13.3%) and 2019 (n = 24; 9.1%) with a gradual decrease in the last two years. The trend was similar for the higher number of EGFRi-related reports: from 2006 to 2017 most of the reports concerned ERL (n = 132; 50.2%), from 2018 to 2020 AFT (n =22; 8.4%), while in the last year OSI (n =5; 1.9%) ( Figure 1 ). Reports having as suspected drug BRG or LORL were not detected in the Sicilian database. As reported in national data, not serious ADRs occurred more frequently (not serious, 73.0% vs. serious, 25.9%). Focusing on gender, a higher number of reports was noted for females except for the EGFRi AFT and GEF that were mostly reported in males (n = 29; 69.0% and n = 33; 71.7%, respectively). Considering all ADR reports, the age group >65 years was the most described except for ALEC, CER, CRIZ, NTB, and OSI that were mainly associated with the age group 18-65 years ( Table 3 ).

Sicilian TKI-related ADRs were similar to national reports, although other relevant ADRs were the onset of xerosis and decreased appetite with ERL (n = 12; 8.7% and n = 6; 4.3%, respectively) and the onset of bradycardia with CER (n = 2; 66.7%) ( Table 4 ). The case-by-case assessment of serious Sicilian ADR reports is described in Table 5 . Considering all reports with only one TKI reported as suspected, the median TTO of ADRs was almost equal for all EGFRi and ALKi (from 37 to 88 days) but it was slightly lower for NTB which was related to a median (Q1-Q3) TTO of 25 (6.5-44) days ( Figure 2 ). Out of the 263 collected reports, SADRs that led to hospitalization, life-threatening, persistent or significant disability, and death were 33 (12.5%) with a median TTO of 57.5 (32.3-135.5) days and mainly involved ERL (n = 17; 51.5%), followed by cases related to GEF (n = 7; 21.2%), OSI (n = 3; 9.1%), CRIZ (n =2; 6.1%), and AFT (n =1; 3%). Only 2 SADR reports had as suspected drugs a combination of a TKI and other agents approved for NSCLC. The majority of reports involved hospitalization (n = 24; 72.7%), followed by death (n = 5; 15.2%), life-threatening events (n = 3; 9.1%), and persistent or significant disabilities (n = 1; 3%). The causality assessment showed that ADRs were mainly possible (60.6%). SADR reports occurred mostly in males (males, 63.6% vs. females, 36.4%) and the median age was of 64 (57-69) years. Reports mostly described a higher percentage of respiratory disorders (n = 10; 30.3%), including respiratory failure, interstitial lung disease, and dyspnea that occurred in men and were mainly related to ERL (n = 9; 90.0%). Four cases of death related to ERL were associated with the onset of epilepsy and cerebral ischemia in a man dead after 49 days, the occurrence of pneumonia with neutropenia and leukopenia in a man dead after 26 days, the onset of respiratory failure and cardiogenic shock in a man dead after 7 days, and the occurrence of intestinal lung disease in a man of 37 years that died after 8 days. Two cases of death were associated with OSI and concerned a case of neoplasm progression in a women of 55 years and a case of sudden death in a man of 48 years.

This is the first European study focusing on ADRs related to the use of TKIs approved for the treatment of NSCLC based on a spontaneous reporting database. Spontaneous reporting System (SRS) is one of the most commonly used pharmacovigilance methods, particularly useful for identifying newly approved drug-related ADRs undetected during pre-marketing studies when the drug is administered to a relatively small number of selected patients and under controlled conditions (60–62). Information coming from multiple ICSRs are a fundamental tool to identify potential safety signals. The use of SRS data has several limitations, including underreporting of suspected ADRs, selective reporting, lack of denominator data (i.e., the total number of patients who have undergone treatment with TKIs), which prevent measuring the absolute risk of suspected ADRs; moreover, SRS relies entirely on individuals’ motivation to report suspected ADRs to a local or national pharmacovigilance center (63). TKIs are mainly used in the advanced stage of NSCLC, in case of relapses or as second-line therapy after treatment with conventional chemotherapy. Consequently, it cannot be excluded that some of the reported ADRs, including cases of death, may be due to the neoplasm progression, to the neoplasm staging, to the onset of delayed ADRs or to some comorbidities in cancer patients treated with TKIs (29). In addition, Naranjo’s algorithm for causality assessment also has limitations such as lack of sensitivity because many responses are classified as unknown in the absence of data that does not always allow the evaluation of possible alternative causes. It also has a serious disadvantage in verifying or invalidating the causality and, at the same time, is not able to provide accurate information on the quantitative extent of the probability assessment (64). Furthermore, the MedDRA PT classification has some limitations, in particular to identify the correct PT that can be reported with different PT synonyms for the same clinical condition, which leads to the false representation of reported ADR (65). A detailed description of ADR reports is available only for Sicilian data collected into the Italian SRS database and recently approved drugs, including BRG and LORL, were not detected in the Sicilian database. Despite these limitations, it is known that SRS contributes to the characterization of safety profiles, which is particularly important for preventing some ADRs related to the use of TKIs in the treatment of NSCLC.

The importance of the SRS database for the identification of TKI-induced ADRs for the treatment of NSCLC was shown in the present study. The analysis of spontaneous ADR reports of TKIs confirmed, in general, well-known risks associated to these drugs in accordance with the frequency of AEs described in each Summary of Product Characteristic, which often include skin, gastrointestinal, general, liver, and respiratory diseases as well as infections. All TKIs are still better tolerated than chemotherapy and generally have better therapeutic results. Our findings were a fundamental tool to help the oncologists in the management of ADRs in clinical practice, especially for some kind of toxicity. Furthermore, more attention should be paid to the occurrence of serious ADRs including respiratory failure, interstitial lung disease, and cardiogenic shock as they are the ones inducing to death even in young patients. More analyses should be conducted to get further information on the safety profile of TKIs considering their effectiveness in treating different types of cancer, but also for the onset of ADRs that can reduce therapy compliance and should be recognized and managed as soon as possible. Close collaboration between oncologists and pharmacologists is required to early identify ADRs, make patients aware of the SRS and thus incentivize reporting to reduce risks appropriately.