ACE: Angiotensin-Converting Enzyme Inhibitors; BAL: Bronchoscopy with Bronchoalveolar Lavage; CT: Chest Tomography; COVID-19: Coronavirus Disease of 2019; DAD: Diffuse Alvelaor Damage; DILD: Drug Induced Lung Disease; DI-ILD: Drug-Induced Interstitial Lung Disease; DLCO: Diffusing Capacity of the Lung for Carbon Monoxide; DMARDs: Disease-Modifying Anti-Rheumatic Drugs; EP: Eosinophilic Pneumonia; GGO: Ground-Glass Opacity; HP: Hypersensitivity Pneumonitis; ILD: Interstitial Lung Disease; LEF: Leflunomide; MTX: Methotrexate; NSIP: Nonspecific Interstitial Pneumonia; OP: Organizing Pneumonia; PFT: Pulmonary Function Testing; SR: Sarcoidis like Reaction; TNF-a: Tumor Necrosis Factor Alpha; UIP: Usual Interstitial Pneumonia

Drug-induced lung disease (DILD) can involve various lung parts, including the airways, lung parenchyma, pleura, vasculature, and neuromuscular system [1]. The most prevalent form of DILD is interstitial lung disease (ILD), characterized by varying levels of inflammation and fibrosis within the lung parenchyma [1,2]. The global incidence of ILD remains unclear, but 2.5-3% of cases are drug-induced [1]. Drug-induced interstitial lung disease (DI-ILD) encompasses a heterogeneous group of manifestations, ranging from mild to severe, life-threatening conditions. More than 400 drugs have been implicated in causing ILD, including biological agents, cardiovascular medications, antibiotics and chemotherapy [3]. This list is expected to grow as new therapeutic agents are introduced. Recognition of DI-ILD remains challenging due its nonspecific clinical, laboratory, radiological and histological features, which vary widely both among different drugs and between individuals using the same drug [3]. The diagnosis of DI-ILD primarily depends on identifying exposure to a drug known for lung toxicity and ruling out other potential causes. DI-ILD can significantly impact patients globally, influencing survival across various contexts. However, the true incidence is likely still underestimated due to inconsistent diagnostic criteria and underreporting [4].

This study aims to analyze the epidemiological and clinical characteristics of patients with DI- ILD and identify the implicated drugs, based on cases reported to the National Pharmacovigilance Center of Tunisia.

A descriptive retrospective study was performed. We included all cases of DI-ILD; notified to the National Center for Pharmacovigilance (CNPV) of Tunisia during a 10-year period from October 2013 to September 2023.

After excluding patients with undefined diagnoses and those without chest tomography (CT) available, we finalized our study population. Demographics, complete clinical history, CT and further investigations were collected for each patient.

The drug causality assessment was established according to the Naranjo’s Adverse Drug Reaction (ADR) Probability Scale [5]. This ADR Probability Scale consists of 10 questions, each answered with "Yes", "No", or "Do not know", and assigned point values from -1 to +2 based on the response. The questions evaluate various aspects of the adverse event, such as its timing relative to drug administration, improvement upon drug discontinuation or antagonist use, recurrence upon drug readministration, alternative causes, and dose-response relationships. Other criteria include the presence of toxic drug concentrations in biological samples, recurrence upon placebo administration, past reactions to similar drugs, and confirmation through objective evidence. Scores are tailed ranging from -4 to +13, to categorize the probability of causation:

• 9 or higher: indicates definite
• 5 to 8: suggests probable
• 1 o 4: is possible
• 0: doubtful

Various diagnoses were reported, all classified as ILD based on the criteria established by the American Thoracic Society/European Respiratory Society (ATS/ERS) [6]. These included:

• Nonspecific Interstitial Pneumonia (NSIP): characterized by ground glass opacities (GGO) predominantly in the lower lobes, accompanied by varying degrees of reticular changes and traction bronchiectasis.
• Organizing pneumonia (OP): marked by inflammatory infiltrates in the respiratory bronchioles and alveoli, with patchy airspace consolidations typically distributed peripherally.
• Hypersensitivity pneumonitis (HP): defined by interstitial lymphocyte infiltration, epithelial cell hyperplasia, and the formation of small granulomas.
• Sarcoidis like reaction (SR): presents with hilar and mediastinal lymphadenopathy along with peribroncho-vascular thickening.
• Eosinophilic pneumonia (EP): features areas of consolidation and GGO, predominantly in the lower lobes, with presence of eosinophilic infiltration in the pulmonary interstitium.
• Usual interstitial pneumonia (UIP): identified by honeycombing with a predominantly basal and subpleural distribution, diffuse traction bronchiectasis, and intralobular reticulations.

During our study period, from October 2013 to September 2023, a total of 35,172 adverse drug reactions (ADRs) were reported to the CNPV. Among these, we identified 42 cases potentially corresponding to DI-ILD. Inclusion and exclusion criteria were applied to these cases, resulting in a final study population of 20 patients. The identified cases of DI-ILD accounted for 0.0006% of all notifications received by the CNPV during the study period.

Patients’ characteristics

Of the 20 patients, 11 were female (sex-ratio F/M=0.55), with a median age of 59.9 years (range: 22-75). Three patients had personal atopy, including two with allergic rhinitis and one with a food allergy. None had a history of drug hypersensitivity.

Four patients had underlying lung diseases: pulmonary tuberculosis (n=2), sarcoidosis (n=1), and chronic obstructive pulmonary disease (n=1). Three patients diagnosed since 2020 had a history of COVID-19 infection within the previous year. Eight patients were smokers, two consumed alcohol, and five had high-risk occupational exposure, including masonry (n=1), welding (n=1), poultry farming (n=2) and industrial plastics manufacturing (n=1). Comorbidities were found in 16 patients including chronic inflammatory diseases (n=8), such as rheumatoid arthritis (n=4), ulcerative colitis (n=4), and psoriatic arthritis (n=2). Other conditions included hypertension (n=6), heart diseases (n=5) and diabetes (n=2).

Clinical presentation

The median time to symptom onset was 30 months (range: 1 month to 20 years). Symptoms appeared in 70% of cases within one year of drug initiation. Respiratory symptoms predominantly included cough in 18 cases and dyspnea in 5 cases. These symptoms were categorized as grade 2 (isolated dry cough or effort dyspnea), with no grade 3 or 4 manifestations reported. Lung auscultation was unremarkable in 17 out of 20 cases, while crackles and wheezing were observed in the remaining three cases.

Paraclinical investigations

Chest radiography was performed for all cases, revealing no specific abnormalities or pleural effusions. Diagnosis was confirmed in all cases based on characteristic findings on CT scans. The radiological patterns observed included NSIP (n=9), OP (n=3), HP (n=3), SR (n=2), EP (n=2) and UIP (n=1). Fibrosis was identified in 12 out of 20 patients; 10 exhibited advanced fibrosis, while the remaining two patients had early, mild fibrosis. Blood tests revealed mild eosinophilia in two cases, with no significant abnormalities in the others. Bronchoscopy with bronchoalveolar lavage (BAL) was performed in four cases, detecting inflammatory cytology in two. Pulmonary function testing (PFT) was also conducted in four cases, with only one case showing mild gas exchange impairement without restrictive or obstructive ventilator defects. None of the patients underwent lung biopsy.

Potential triggers

The suspected causative agents in our cases included methotrexate (MTX) (n=5), infliximab (n=2), fluoxetine (n=3), sertraline (n=1), amiodarone (n=1), captopril (n=1), atorvastatin (n=1), pravastatin (n=1), sulfasalazine (n=1), mesalazine (n=1), teriflunomide (n=1), cyclophosphamide (n=1) and peginterferon (n=1).

Management and outcomes

In nine cases, the suspected agent was continued. Among these, four patients experienced persistent symptoms without exacerbation during a two-year follow-up, while the remaining five patients showed worsening symptoms and progression of radiological lesions. The causative drug was discontinued in 11 cases. Of these, four patients showed clinical improvement with symptom regression within 10 to 30 days under corticosteroid therapy, accompanied by stable radiological findings. Complete remission of both symptoms and imaging abnormalities occurred in one case within four months. However, three patients continued to experience respiratory symptoms despite drug withdrawal during a follow-up period of three months to one year. Outcomes for the remaining four patients were not documented.

Imputation scores

In our study, the imputation scores were classified as either possible or probable, with no cases reaching the definite category. All cases were initially assigned a baseline score of 1, as ILD had been previously reported in association with all the suspected drugs.

A score of 5 was attributed to patients in whom ILD developed after initiating the suspected drug (+2), coupled with absence of alternative causes (+2). In two cases, the score increased to 6 due to objective evidence of the adverse event (case 10) or clinical improvement following drug discontinuation (case 13). A score of 2 was assigned to cases where the delay of onset was compatible with drug-induced ILD, but other potential causes, primarily chronic inflammatory diseases, were present (-1). In cases where drug discontinuation led to a favorable clinical outcome, the score was increased to 3 (cases 12, 17, and 18) (Table 1).

In this retrospective study conducted over a 10-year period, a total of 20 cases of DI-ILD were identified. The sample size of 20 cases collected over at a decade in a country the size of Tunisia is limited and is most likely representing under reporting rate within Tunisian pharmacovigilance systems, compounded by the inherent challenges in diagnosing this condition. Pharmacovigilance in Tunisia has historically faced limitations, particularly in the collection and interpretation of rare or complex adverse drug reactions such as DI-ILD. Healthcare professionals may not consistently report cases due to time constraints or the difficulty of definitively linking lung injury to drug exposure.

In line with these findings, several studies suggest that DI-ILD is underdiagnosed worldwide. The global incidence remains unclear [1], with reported rates varying significantly among drugs, ranging from less than 1% to nearly 60% [4]. DI-ILD accounts for approximately 3% to 5% of of all ILD cases, corresponding to an estimated incidence of 4.1 to 12.4 cases per million per year [3]. However, these figures likely underestimate the true burden, especially given the increasing use of novel oncology drugs associated with high DI-ILD rates. This underscores the urgent need for improved awareness, robust reporting mechanisms, and enhanced diagnostic capabilities to better capture the true incidence of DI-ILD. Addressing these gaps is essential to promoting patient safety and ensures appropriate management of drug-related toxicities.

In our series, the onset of symptoms ranged from 1 month to 20 years, with a median delay of 2.5 years. Similarly, the literature reports a broad variation in the time to onset of DI-ILD, spanning from just a few days to several months, and in some cases exceeding 15 years following the initiation of treatment [3,7].

In our series, DI-ILD manifested as dry cough and/or dyspnea in all cases. However, clinical presentations reported in the literature vary widely, ranging from asymptomatic to severely symptomatic patients [8]. Acute pneumonitis typically manifests with shortness of breath and fever, while chronic onset is characterized by progressive dyspnea and cough [3].

In DI-ILD patients, chest auscultation often reveals fine crackles as reported in the literature [2,3]. However, in our series, crackles were detected in only three cases. While lung crackles are commonly associated with ILD, they can be not audible in early or mild stages of the disease. Additionally, the detection of crackles can be influenced by various factors including the examiner’s technique, the patient’s positioning, and the extent of pulmonary involvement. In cases of localized or minimally pronounced lung involvement, crackles might not be perceptible.

Chest radiographs are normal in 25 to 75% of DI-ILD cases. However, in other instances, pulmonary infiltrates may represent the earliest radiological sign of DI-ILD [4,9]. Notably, no pleural effusions were observed in our series.

CT imaging is the preferred modality for diagnosing ILD [4]. The most commonly reported disease pattern is nonspecific interstitial pneumonia, followed by organizing pneumonia and hypersensitivity pneumonitis, which aligns with our findings [4,8]. However, CT scans are rarely specific in identifying a drug related etiology [1], as many drugs can present with multiple patterns, and conversely, a single pattern can be associated with different drugs [4,8]. For instance, among our five MTX cases, three distinct patterns were observed: NSIP (n=3), SR (n=1), and HP (n=1).

Laboratory tests are generally nonspecific but can be useful in ruling out alternative causes of ILD. A hemogram, for instance, may reveal an elevated eosinophil count in cases of EP and, less commonly, in HP cases [3]. In our series, both patients diagnosed with EP demonstrated increased eosinophil count, whereas complete blood count for patients with HP showed no abnormalities. These findings highlight the potential role of laboratory evaluations in differentiating between different patterns of ILD, even though their overall diagnostic specificity remains limited.

Bronchoscopy is a crucial tool in the diagnostic work-up for ILD, mainly to exclude alternative causes, such as infection and malignancies [3]. While BAL findings, including specific or mixed cellular patterns, are often observed, they are not exclusive to DI-ILD and can also occur in other inflammatory or infectious lung diseases [1,4]. The most characteristic BAL feature of DI-ILD is lymphocytic alveolitis with a predominance of CD8⁺ cells [3]. Additional findings reported in DI-ILD include cellular abnormalities, such as nuclear enlargement, hyperchromasia, lipid inclusions, and haemosiderin-laden macrophages [4]. In our series, the two cases of EP exhibited an eosinophilic cellular pattern in the BAL fluid, which is typical for this condition [1].

PFTs are crucial for the diagnosis and follow-up of patients with ILD, typically revealing a restrictive ventilatory defect with impaired gas exchange, indicated by a reduced DLCO (diffusing capacity for carbon monoxide) [3]. However, an obstructive ventilatory defect my also occur in some cases [1]. In our series, PFTs were performed in only four patients, with normal results in three—an unusual finding in ILD.

The limited use of PFTs performed in our cases can be explained by several factors. The decision to perform PFTs often depends on the clinical presentation and symptom severity. For patients with mild or stable symptoms, clinicians may prioritize imaging and symptom monitoring over PFTs, reserving them for cases with significant respiratory dysfunction. Access barriers also contribute to underutilization. Studies frequently highlight limited access to lung function tests as a challenge in diagnosing pulmonary diseases [10]. Tunisian patients face additional obstacles, including travel difficulties, insurance limitations, and insufficient spirometry availability. Hospitals often contend with high patient volumes, resource constraints, and equipment shortages, leading to significant delays in scheduling PFTs. Consequently, clinicians may rely on clinical and radiographic findings rather than waiting for PFT results in cases with strong suspicion of DI-ILD.

In our series, normal lung function cannot be assumed for patients who did not undergo PFTs, as the absence of formal testing leaves this unconfirmed. Among patients with normal PFT results, it is important to note that normal findings can still occur despite histologic and radiographic evidence of ILD. Therefors, normal PFTs do not exclude an ILD diagnosis when clinical or radiographic abnormalities are present [11]. While functional abnormalities are characteristic of ILDs, they lack disease-specificity. Although distinct lung function patterns have been documented in ILDs, their significant overlap limits their diagnostic value in differentiating between specific types of ILD [12].

In our series, no patients underwent histologic assessment. Diagnosing DI-ILD rarely necessitates histological confirmation, as drugs can induce diverse histopathological patterns of interstitial pneumopathy. Furthermore, a single histological pattern may be linked to various drugs [3].

Potential triggers

Cancer drugs are the leading cause of DI-ILD, accounting for 23–51% of cases in the literature. Disease-modifying anti-rheumatic drugs (DMARDs) follow, with frequencies ranging from 6–72%, then antibiotics (6–26%), non-steroidal anti-inflammatory drugs (NSAIDs) (0–23%), psychiatric medications and anti-arrhythmic agents with the same frequency (0–9%) [4]. In our study, DMARDs were implicated in the majority of cases (n=8), followed by cardiovascular drugs (n=4), psychiatric medications (n=4) and anti-inflammatory drugs (n=2). In contrast, chemotherapeutic agents were identified in only one case, and no cases involving antibiotics were observed. Common causative drugs for ILD include MTX, amiodarone, and bleomycin [8]. Among our cases, MTX was the most frequently implicated agent, while amiodarone was suspected in only one case.

Cancer drugs

Lung toxicity occurs in 10-20% of patients treated with antineoplastic agents [3]. Bleomycin and cyclophosphamide are commonly implicated drugs [1]. Although chemotherapeutic agents are significant contributors to ILD, only one case of cyclophosphamide-induced ILD was noted in our series. Cyclophosphamide is associated with ILD in approximately 1% of cases when used as a single agent [13]. Onset typically occurs within one to six months of treatment initiation, as observed in our case, but late presentations after years of low-dose therapy have also been reported [13]. Chemotherapy-induced ILD often has a poor prognosis, with severe outcomes even after drug discontinuation [3]. In our patient, cyclophosphamide treatment spanned 18 months, leading to worsening symptoms and radiological abnormalities.

Cardiovascular drugs

Cardiovascular medications associated with ILD include antiarrhythmics, angiotensin-converting enzyme inhibitors (ACE), anticoagulants, and statins [8]. Our series included one case involving amiodarone, one with captopril and two with statins (atorvastatin and rosuvastatin).

Amiodarone

Amiodarone is the most common cardiovascular drug linked to ILD, affecting up to 5% of treated patients, with fatality rates of 10-20% [1,3]. ILD usually develops within 6-12 months of treatment but can occur earlier or after years of therapy [3]. In our case, the onset was two years after treatment initiation. PFTs showed reduced DLCO without restrictive defects, which is atypical [3]. The characteristic “foamy” macrophages were absent [3]. Despite the long half-life of amiodarone and its potential for disease progression post-discontinuation [3], our patient tolerated the drug for two years without exacerbation. Amiodarone is associated with an EP pattern [8], which was consistent with our findings. Risk factors include doses exceeding 400 mg/day, long-term use and age >60 years [3]. However, our patient, despite being 60 years old, received a low dose (200 mg/day) and had no pre-existing lung disease. Pathogenic mechanisms include direct toxicity, immune reactions, and disruption of thyroid hormone signaling [3,8].

Statins

Statin-induced ILD is considered a class effect rather than drug-specific [1]. Atorvastatin is more frequently associated with ILD than other statins [14]. Onset varies from months to years, with improvement often observed after discontinuation [15]. Among our cases, one patient developed ILD after two years of pravastatin use, with no symptom exacerbation during a three-month follow-up. The second patient presented ILD shortly after starting atorvastatin, with symptoms resolving within one month of drug cessation. Potential mechanisms include phospholipase inhibition as part of statins’ effects on lipid metabolism [14].

Captopril

Captopril-induced ILD is rare most commonly associated with EP pattern [16]. In contrast, our case developed HP 18 months after treatment initiation.

Disease-Modifying antirheumatic drugs (DMARDs)

DI-ILD is a rare but severe adverse reaction with nearly all DMARDs [17]. In our series, eight cases involved DMARDs.

Methotrexate

MTX-induced ILD is estimated to occur in 0.3%-8% of patients with rheumatic disorders [17]. Onset ranges from weeks to months after therapy initiation, but cases have been reported following short-term, high dose regimens [3]. Among our cases, four developed ILD after more than two years of treatment, while one occurred after one month on 15 mg/week. Patterns included NSIP (n=3), OP (n=1) and HP (n=1). Fibrosis was rare, noted in only one case. Cessation of MTX may be sufficient for clinical improvement and even disease reversal to occur. Risk factors such as advanced age, pre-existing lung disease, and diabetes were absent except for age>60 years in two cases [3].

Infliximab and leflunomide (LEF)

Infliximab-induced ILD often occurs within 6-7 weeks of treatment initiation, as seen in one of our cases [17]. LEF-induced ILD, with a prevalence of 0.02%, is associated with significant mortality (20%) and frequently occurs within 20 weeks of treatment initiation [18]. In our case, ILD was delayed by two years. Despite discontinuation, symptoms persisted, highlighting LEF’s long half-life and slow resolution [18].

Anti-Inflammatory drugs

Sulfasalazine, metabolized into mesalazine and sulfapyridine, is widely known for hypersensitivity reactions attributed to its sulfapyridine component [19]. Mesalazine, introduced as a safe alternative, has been associated with increasing reports of pulmonary toxicity, albeit unfrequently [20]. In our study, two patients developed ILD following long-term sulfasalazine (20 years) and mesalazine (2 years) use, respectively. Notably, one patient fully recovered within four months after discontinuing mesalazine, while the second experienced persistent symptoms for one year after sulfasalazine withdrawal, likely due to delayed diagnosis. The pathogenesis of mesalazine-induced ILD remains unclear, with potential mechanisms including immune-mediated alveolitis or direct toxicity to alveolar structures [19].

Peginterferon α-2b

Interferon-induced pulmonary toxicity is rare and typically occurs within 2-12 weeks of treatment initiation [21]. In our patient, ILD emerged eight weeks after starting peginterferon α-2b monotherapy for chronic hepatitis C. Although most reported cases resolve after drug cessation [22], our patient exhibited persistent respiratory symptoms five years post-withdrawal. Peoposed mechanisms include interferon-induced immune responses, increased endothelin-1 expression, and upregulation of pulmonary histocompatibility antigens [23].

Psychiatric medications

Lungs can act as reservoirs for selective serotonin reuptake inhibitors (SSRIs) [24], contributing to rare cases of ILD. Four cases were observed in our study: three involving fluoxetine and one sertraline.

Fluoxetine

In our series, ILD manifested five years after initiating fluoxetine. Literature reports describe variable exposure durations, typically a few months to over 20 years [24]. While drug cessation generally improves symptoms, delayed diagnosis may hinder oucomes, as seen in one of our cases. BAL findings in our cases were non-specific. However, literature frequently reported lymphocytic or eosinophilic patterns [24].

Sertraline

Sertraline-induced ILD remains rare, with onset ranging from 7 days to 8 years [25]. In our patient, ILD developed five years post-initiation, with cessation of treatment failing to yield clear clinical outcomes. Proposed mechanisms involve serotonin’s effects on immune cells and potential immune mediated or cytotoxic pathways as observed with other SSRIs like fluoxetine [25].

Management and outcomes

Management of DI-ILD remains challenging due to the lack of robust evidence. Some studies report continuation of the suspected agent even in patients with grade 3 ILD continued [4]. Treatment decisions require careful consideration of the risk of DI-ILD progression versus the potential impact of drug discontinuation on overall survival. Neverthless, whenever possible, discontinuing the causative drug remains the cornestone of management [3]. For patients with severe or progressive disease despite drug withdrawal, glucocorticoids are commonly used, although there are no established guidelines fro their use in DI-ILD [3,4]. The dose and duration of glucocorticoid therapy vary widely across studies. Treatment responses are highly variable, ranging from minimal improvement, especially in patients with diffuse alveolar damage (DAD), to complete resolution in those with OP [3]. In cases of pulmonary fibrosis, corticosteroids are generally ineffective, and disease progression can continue even after discontinuation of the suspected agent [1].

In our series, management approaches and outcomes varied significantly, with only four patients receiving glucocorticoids. In nine cases, the suspected drug was continued, leading to stable symptoms in four patients, while the other five experienced worsening symptoms. Among the 11 patients who had their treatment discontinued, only one achieved remission within four months. This variability in patient responses to drug continuation versus discontinuation aligns with findings in the literature. The outcome of DI-ILD depends on both the drug involved and the ILD pattern. While complete recovery is possible with drug withdrawal, dose reduction, and/or glucocorticoïd use, many patients fail to improve, or experience a progressive course. Chronic symptoms may take longer to resolve, and in some cases, the disease can worsen even after the trigger drug is stopped, potentially leading to respiratory failure and death [4].

Rechallenge with the suspected drug after DI-ILD requires a careful, case-by-case evaluation, considering the severity of the initial reaction and the availability of alternative therapies [3]. Approximately one-third of re-challenged patients experience recurrence of respiratory symptoms, although successful rechallenge after remission has been reported [3]. None of the patients in our series who discontinued treatment was rechallenged.

The prognosis of DI-ILD varies with the trigger drug and disease severity. Common complications include pulmonary fibrosis and respiratory failure [1], with the latter associated with mortality rates exceeding 60% when mechanical ventilation is required [4]. Poor prognostic factors include advanced age (≥ 65 years), smoking, acute and severe disease presentations, reduced lung function, DAD patterns, and pre-existing interstitial pneumonia [1,4].

Risk factors

Certain risk factors for the development of DI-ILD are consistently observed across studies with some also being associated with specific drugs [4].

Increased age has been identified as a significant risk factor for DI-ILD in patients treated with bleomycin, gemcitabine, LEF, MTX, amiodarone and nitrofurantoin [4]. In our series, the eldery was the most affected, regardless of the drug used.

While there is no conclusive evidence suggesting that gender influences the risk of DI-ILD [1], some studies have indicated that male sex may be a risk factor for DI-ILD, particularly following treatment with MTX and amiodarone [4]. In our series, the patient who received amiodarone was male, while the majority of those treated with MTX (4/5) were female.

Pre-existing ILD has been associated with a threefold increased risk of DI-ILD in patients treated with gefitinib, erlotinib, or LEF but not with amiodarone [4,3]. Interestingly, both patients treated with LEF or amiodarone in our series have no underlying lung disease.

Smokers are at higher risk of developing DI-ILD when treated with gemcitabine, tyrosine kinase inhibitors, and MTX [4]. However, none of the patients treated with MTX in our series were smokers.

A clear dose-dependent relationship has been well-established for bleomycin, amiodarone, and nitrofurantoin [4]. In our case, the patient treated with amiodarone received the standard dose of 200 mg per day.

In rheumatoid arthritis, prior MTX exposure has been found to increase the risk of ILD [4]. One of our patients, who developed ILD during infliximab therapy, had a history of prior MTX treatment.

Other potential risk factors for DI-ILD include genetic predispositions, excessive alcohol consumption, impaired renal function, and diabetes [3,4]. Diabetes was observed in two of our cases.

Additionally, a seasonal distribution has been reported as a risk factor for DI-ILD, suggesting that viral infection may act as a cofactor in the development of lung toxicity [3], which was consistent with our findings.

Emerging biomarkers in DI-ILD

Biomarker tests have gained importance in drug development and clinical practice due to their minimally invasive nature and low cost. In DI-ILD, biomarkers such as Krebs von den Lungen-6 (KL-6) and surfactant proteins (SP-A and SP-D)— glycoproteins produced by type II pneumocytes— are commonly elevated and have been investigated for diagnostic purposes. KL-6 is considered a sensitive marker for certain ILD patterns, including DAD and NSIP, while SPs are more specific to pulmonary fibrosis. ADAM8 (a disintegrin and metalloproteinase 8) has also been reported to be elevated in drug-induced EP. However, these markers lack specificity, as they may also be elevated in other pulmonary conditions, such as lung cancer, idiopathic interstitial pneumonia, and connective tissue disease-related ILD, thereby limiting their diagnostic value [26-28].

To address this gap, omics-based approaches, particularly lipidomics, have identified lysophosphatidylcholines (LPCs) as promising biomarkers for DI-ILD. Amng these, LPC(14:0) outperformed KL-6 and SPs in distinguishing DI-ILD from other lung diseases. Unlike classical markers, LPC(14:0) was not affected by the specific causal drug, but correlated with disease severity. Thus, LPC(14:0) may serve as a novel, sensitive, and specific biomarker to enhance early diagnosis and management of DI-ILD [26]. 

In addition to these emerging biomarkers, interest is growing in the role of molecular regulators such as the sitruin family. SIRT1 is an NAD⁺-dependent deacetylase involved in regulating inflammation and stress response. Its expression is reduced in age-related lung diseases such as chronic obstructive pulmonary disease, lung cancer, and idiopathic pulmonary fibrosis. Although direct evidence linking SIRT1 to DI-ILD is limited, related studies suggest the antifibrotic potential within the human sirtuin family [29]. Notably, SIRT6 has been shown to protect against bleomycin-induced lung injury by promoting lipid degradation through the PPARα activation, thereby reducing inflammation and fibrosis. These findings warrant further investigation into SIRT1 and its regulatory roles as potential biomarkers or therapeutic targets in DI-ILD [30].

This study has several strengths, including being one of the rare investigations into ILD in Tunisia, with a long study period that provides valuable insights over time. It offers a comprehensive analysis of all ILD characteristics, detailing clinical information and contributing relevant findings to the existing literature. Additionally, a thorough review of key ILD features and potential triggers was conducted. However, there are notable limitations. The retrospective nature of the study restricts the ability to identify specific risk factors for ILD, and many if not most of the cases did not have a proper clinical workup like lung-function test and BAL making the diagnosis of drug-induced lung disease questionable. Incomplete investigations and follow-ups in some cases further reduced the comprehensiveness of the study. Although the study includes a summary of literature supporting that the drugs incriminated in this study actually can case drug-induced lung diseases, this does not constitue definitive evidence for causality in the 20 reported cases. The small number of cases, despite the long duration of the study, limits the generalizability of the findings, while the presence of multiple underlying diseases and the absence of confirmatory histology led to a low imputability score and a lack of definitive cases, preventing full conclusions about the responsibility of suspected drugs. Based on these findings, we recommend that future research focus on larger prospective studies with standardized diagnostic tests and comprehensive clinical workups to confirm the role of drugs in ILD development. Increasing the sample size, ensuring detailed histological examination, and including confirmatory evidence will help better understand the relationship between drug exposure and ILD. Additionally, improving data collection on clinical characteristics will enhance the identification of risk factors and triggers.

This study presents an analysis of 20 cases of DI-ILD reported to the National Center of Pharmacovigilance of Tunisia, highlighting their epidemiological and clinical characteristics, as well as the implicated drugs. Our findings are largely consistent with previous reviews, although several limitations should be acknowledged. The main conclusion is that DI-ILD remains severely underreported in Tunisia, as in many other countries, and that patients suspected of having this condition often do not undergo appropriate clinical diagnostic procedures to confirm the diagnosis. DI-ILD represents a highly heterogeneous group of conditions with an expanding list of potential triggers. Establishing a definitive causal relationship remains challenging and requires thorough investigation. Large prospective studies are essential to develop clear diagnostic strategies and evidence-based management guidelines.

YSM: data collection, literature review, conceptualization, design of the study, data analysis, interpretation of the results, and writing the manuscript; ID: data collection, and assisted in the analysis and interpretation of the findings; FZ: design of the study and assisted in the critical revision of the manuscript; IA: provided methodological support and critically revised the manuscript; SEA: provided overall supervision of the study, contributed to data interpretation, and critically reviewed the manuscript.

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

Given the retrospective nature of the study, written consent was not required. However, oral consent was obtained from all patients, and their anonymity was strictly preserved in accordance with ethical guidelines.

This research received no funding.