Drug-induced pulmonary arterial hypertension: a primer for clinicians and scientists

Abstract

Drug-induced pulmonary arterial hypertension (D-PAH) is a form of World Health Organization Group 1 pulmonary hypertension (PH) defined by severe small vessel loss and obstructive vasculopathy, which leads to progressive right heart failure and death. To date, 16 different compounds have been associated with D-PAH, including anorexigens, recreational stimulants, and more recently, several Food and Drug Administration-approved medications. Although the clinical manifestation, pathology, and hemodynamic profile of D-PAH are indistinguishable from other forms of pulmonary arterial hypertension, its clinical course can be unpredictable and to some degree dependent on removal of the offending agent. Because only a subset of individuals develop D-PAH, it is probable that genetic susceptibilities play a role in the pathogenesis, but the characterization of the genetic factors responsible for these susceptibilities remains rudimentary. Besides aggressive treatment with PH-specific therapies, the major challenge in the management of D-PAH remains the early identification of compounds capable of injuring the pulmonary circulation in susceptible individuals. The implementation of pharmacovigilance, precision medicine strategies, and global warning systems will help facilitate the identification of high-risk drugs and incentivize regulatory strategies to prevent further outbreaks of D-PAH. The goal for this review is to inform clinicians and scientists of the prevalence of D-PAH and to highlight the growing number of common drugs that have been associated with the disease.

FIFTY YEARS OF DRUG-INDUCED PULMONARY ARTERIAL HYPERTENSION: A HISTORICAL PERSPECTIVE

Until 1968, primary pulmonary hypertension (PPH) was a poorly characterized disease because of the challenges of studying pulmonary vascular physiology and the low prevalence of the disease. Despite the fact that the pathology had already been documented, the physiology of PH was unknown until the advent of right heart catheterization (RHC) in the 1950s (11). However, in 1968, international interest in further characterizing PH increased because of a rapid rise in the number of cases of PPH in Switzerland, Germany, and Austria (58, 59).

Three years prior, aminorex fumarate (sold under the name Menocil), an amphetamine-like appetite suppressant, had come on the market as a novel treatment for obesity (Fig. 1). In the years that followed, an increase in the incidence of PPH was noted. One or two in 1,000 aminorex users developed PH, a significantly higher incidence rate than primary pulmonary arterial hypertension (PAH, estimated at 1–2 cases per million per year). For almost 60% of patients diagnosed within this time frame, symptoms began ~6 mo after the initiation of aminorex intake, although some noted symptoms within weeks of starting to use the drug. While several patients had regression of their disease after discontinuing aminorex, the mortality rate was high, with nearly one-half of the aminorex-exposed PPH patients dying within 10 years of diagnosis from pulmonary hypertension (PH)-related complications (53). Aminorex was taken off the market in Europe in 1972 following these case reports but remained on the market in the United States (U.S.) until 1973.

Fig. 1.

Molecular structure of amphetamine-derived anorexinogens.

In response to the aminorex-associated PPH epidemic, the World Health Organization (WHO) organized the First WHO Pulmonary Hypertension Conference in 1973 to assess the current state of knowledge and define the nomenclature and classification system that would be used to describe PH in the future. The conference led to a commitment to better understand the nature of normal pulmonary circulation and to carefully study cases of PH (62). In the U.S., the National Heart, Lung, and Blood Institute of the National Institutes of Health created a national database of PH cases shortly after the First WHO Pulmonary Hypertension Conference. Within just a few years, nearly 200 patients had been added to the database, and these early patients’ data contributed significantly to the evolving understanding of the pathophysiology of PH (125).

Around the same time, cases of PH were associated with other drugs and substances, raising further concern that PH could be triggered by substance exposure. Toxic oil syndrome was a scleroderma-like disease with vascular endothelial swelling and proliferation associated with PH triggered by exposure to a toxic rapeseed oil anilin derivative (50). Approximately 20,000 cases of PH associated with toxic oil syndrome were identified, resulting in >300 deaths (80). In 1989, another outbreak of PH, described as eosinophilia-myalgia syndrome, occurred in Mexico in association with a toxic product of l-tryptophan (64). In many of these cases, PH developed in patients who seemed to have an underlying genetic predisposition that was unmasked by exposure to the substance (7). In addition, phenformin, one of the early biguanides, was associated with the development of PH in a few patients; the PH resolved upon cessation of the drug (41).

Most notable of the substances associated with PH during this time was fenfluramine, popularly prescribed in combination with phentermine as “Fen-phen,” which became available on the market in the late 1970s. Like aminorex, fenfluramine is an amphetamine-like drug with indirect serotonergic activity used for weight loss. In 1981, the first report of fenfluramine-associated PPH was published (34). The International Primary Pulmonary Hypertension Study (IPPHS), a multicenter case control study conducted between 1992 and 1994, subsequently demonstrated a definite risk of PH associated with anorexic agents like fenfluramine (2). Abenhaim et al. recruited patients diagnosed with primary PH at centers throughout France, Belgium, the United Kingdom, and the Netherlands. Patients with diagnoses associated with secondary PH, such as interstitial lung disease or myocardial disease, were excluded. Age- and gender-matched controls were identified from regional primary care practices. The exposure status was assessed via structured face-to-face interviews by trained blinded interviewers with additional information provided from review of the medical chart. The drugs of interest included fenfluramine and its relative dexfenfluramine, amphetamine-like anorexic agents, and other appetite suppressants. The study included 95 cases and 355 controls. Use of anorexic drugs, predominantly fenfluramine and dexfenfluramine, was associated with an odds ratio of 6.3 (95% confidence interval: 3.0–13.2) of having primary PH, and if those substances had been used for more than 3 mo, the association was even stronger with an odds ratio of 23.1 (6.9–77.7). Tragically, dexfenfluramine was approved in the U.S. for treatment of obesity in 1996. Of note, the editorial that accompanied the New England Journal of Medicine article by Abenhaim et al. concluded that the benefits of these agents outweighed the risks of PAH, thus allowing prescription of the medication to continue in the U.S. (92).

In the years following IPPHS, two additional case-control studies, the Surveillance of North American Pulmonary Hypertension study (126) and the Surveillance of Pulmonary Hypertension in America study (144), confirmed that there was a strong association between fenfluramine exposure and PH. During this same time, evidence mounted that fenfluramine and its relatives were associated with cardiac valvular lesions (23). It was hypothesized that this association was because of the serotonergic effects of fenfluramine and its relatives that led to heart lesions similar to those seen in carcinoid syndrome (96), which interestingly is not associated with pulmonary vascular disease. Shortly after the publication of these reports, in 1997, fenfluramine and dexfenfluramine were taken off the market.

Benfluorex hydrochloride (Mediator) was yet another derivative of fenfluramine marketed as a weight loss and diabetes drug. It was approved in Europe in the 1970s, and, because benfluorex was approved as a treatment for diabetes and metabolic syndrome, rather than as an anorexigen, it did not receive the same level of scrutiny as fenfluramine and dexfenfluramine did in the 1980s and 1990s. As a result, benfluorex was available on the market into the 2000s. However, in 2009 and 2012, two case series were published describing an association between benfluorex exposure and PH (14, 131). Both case series described a combination of PH and valvulopathy, as was seen with fenfluramine and dexfenfluramine, although not all patients with fluramine-induced PH had valve disease. After the first case series in 2009, the European Medicines Agency recommended withdrawal of benfluorex from the market.

RECENT OUTBREAKS OF DRUG-INDUCED PULMONARY ARTERIAL HYPERTENSION ASSOCIATED WITH MEDICATIONS AND TOXINS

Interferons

Interferons (IFN) are cytokines released by immunomodulatory cells in response to infections and malignancies. IFN have been developed as therapy for a variety of immunologic and malignant processes (51). Among these, PEGylated INF-α2a (Pegasys) and -β (Pegintron) are Food and Drug Administration (FDA) approved for the treatment of multiple sclerosis, hematological malignancies, and chronic hepatitis C virus infection. In the last 5 years, case reports have linked IFN use with the development of PAH and raised safety concerns regarding the use of these agents in clinical practice (45, 73, 84). The 2014 French PH registry study reported 53 patients with PAH, 48 of which developed PAH after starting IFN while the remainder had a history of PAH before the initiation of therapy (132). Of the 53 patients, 48 had been treated with IFN-α for a mean duration of 7.7 mo for chronic hepatitis C virus (HCV) (n = 47) and chronic myelogenous leukemia (CML, n = 1). The remaining five patients had been treated with IFN-β for multiple sclerosis, and the time between treatment initiation and diagnosis of PAH was 59–117 mo (i.e., 5–10 yr). Most of these patients developed severe PAH, and two of them died as a result of PAH-related complications. Also, these patients exposed to IFN demonstrated worsening of symptoms and decreased exercise capacity and hemodynamic variables, all of which were parameters reversible in most cases after IFN discontinuation and did not require further treatment with PAH-specific agents. Given the strength of the data accrued so far, IFNs are now recognized as “possible” risk factors for drug-induced pulmonary arterial hypertension (D-PAH) in the most recent 2015 European Society of Cardiology and European Respiratory Society PH guidelines (47).

In an effort to bypass the significant toxicity of IFN for the treatment of HCV, new direct-acting antiviral agents (DAAs) like sofosbuvir have recently become available with demonstrable rates of cure as high as 85% in some studies when combined with ribavirin (49, 83). Whereas use of sofosbuvir and other DAAs has increased in the last 3 years, it must be noted that three cases of severe PAH in HCV patients treated with sofosbuvir-based regimes have been reported (124). Whether PAH in these patients could also be explained by the presence of other risk factors such as HIV and portal hypertension is unknown, but physicians should be aware of this possible D-PAH risk and report cases to pharmacovigilance groups for further monitoring.

Dasatinib

Tyrosine kinase inhibitors (TKIs) are a key treatment for CML, a hematological malignancy caused by chromosomal translocation that leads to the pathogenic tyrosine kinase protein breakpoint cluster region/Abelson (BCR/ABL) (103). TKIs like imatinib (Gleevec) and dasatinib (Sprycel) are being used to treat CML since they can inhibit BCR/ABL kinase activity in CML cells and induce prolonged remission. However, up to 20% of CML patients do not achieve a complete cytogenic response with imatinib. For these patients that do not respond to imatinib, a more potent TKI like dasatinib can be used to induce CML remission. Despite the therapeutic benefit of dasatinib in patients with CML, numerous studies have now shown that dasatinib can lead to PAH development in a subset of these patients.

In 2012, cases were reported from the France PH registry where nine patients had previously been treated with imatinib and subsequently with dasatinib (103). Both drugs were used to treat CML, although dasatinib was prescribed when the disease was found to be nonresponsive to imatinib. At the time of PH diagnosis, the patients presented with progressive dyspnea as well as bilateral pleural effusions in six of nine patients. With the previous evidence of a potential relationship between dasatinib exposure and development of precapillary PH, dasatinib treatment was discontinued in all patients and substituted with the alternate potent TKI nilotinib. In six patients, specific PAH therapy was not necessary, and within 8–36 mo across patients, eight out of nine patients showed clinical and/or hemodynamic improvement. However, despite this improvement, no patient had complete recovery, and two patients died at the time of follow-up. Of these two patients, one did not show any hemodynamic improvement and had the most severe hemodynamic compromise of all nine patients, whereas the other patient experienced sudden death while on a long-distance airplane flight. It must be noted that none of the affected patients had mutations in bone morphogenic protein receptor 2 (BMPR2), that is, mutations of a gene associated with both familial and sporadic forms of PAH.

Since the publication of this report, a study using the Bristol-Myers-Squibb (BMS) international pharmacovigilance database has reported on the clinical course of 41 patients with a diagnosis of dasatinib-induced PAH confirmed by RHC between 2006 and 2013 (134). In this cohort, PAH appeared to develop across a wide timeframe that ranged from as little as 1 mo to almost 7 yr after the initial exposure. Furthermore, cessation of dasatinib led to improvement of PAH in 94% of cases and complete resolution in 58%. Because of the small sample size, the investigators were not able to identify patient characteristics or risk factors associated with dasatinib-induced PAH. Also, a repeat RHC was performed for only five patients, and follow up echocardiograms were performed for only seven patients. Subsequently, Weatherald et al. reported changes in hemodynamics and clinical variables over a period of 1–81 mo (median range 24 mo) in 21 RHC-proven dasatinib-PAH cases from the French Pulmonary Hypertension Registry (146). Similar to the BMS study, most patients demonstrated improvement in hemodynamics, functional class, and 6-min walk distance after cessation of dasatinib. However, more than one-third of patients had persistent PAH when reassessed by RHC during follow-up, and 43% remained symptomatic despite being off dasatinib. The results of this study highlight the possibility that dasatinib can injure the pulmonary vasculature. The authors conclude that patients exposed to dasatinib should receive regular follow-up and initiate therapy if there is clinical and/or hemodynamic evidence of PAH.

To date, no definite cases of PAH have been documented in association with imatinib or nilotinib. It is worth pointing out that, until recently, imatinib had been studied for treatment of PAH, but a phase 3 study failed to demonstrate consistent clinical benefits and raised safety concerns around the drug (67). Of note, cases of PAH associated with bosutinib and ponatinib have also been reported (66, 119), further stressing the need for continued pharmacovigilance over the TKIs.

Mitomycin C

Mitomycin C (MMC) is a bioreductive alkylating agent that is being used as chemotherapy in the management of various malignancies (120). In recent years, MMC has been identified as a possible risk factor for pulmonary veno-occlusive disease (PVOD) (46), a rare form of PAH with an incidence of <0.5 cases per million per year and associated with a very poor prognosis, since most patients die within 2 year of diagnosis (68, 104). Perros et al. described seven cases of MMC-induced PVOD that were identified in the French PH Registry between June 2012 and December 2014 (116). In each case, the primary malignancy was squamous anal cancer, and all patients received MMC, although five of the seven cases also received 5-fluoruracil. The authors note that, in France, treatment with MMC is almost exclusively used in the setting of anal cancer. PVOD was diagnosed in these cases based on clinical, functional, radiological, and hemodynamic changes; a histologic diagnosis was precluded by the tenuous overall clinical condition of patients. Known risk factors for PAH were identified in one patient, who tested positive for human immunodeficiency virus (HIV) infection. The median age of these cases was 53 yr with a predominance of females (6 of 7). At the time of diagnosis, four patients were within New York Heart Association functional class III and three cases in class IV. Even after specific PAH therapies (either as monotherapy or a combination of endothelin-receptor antagonists plus phosphodiesterase type 5 inhibitors or prostacyclins), four of seven patients died: two because of right heart failure, one from local tumor progression, and one patient from severe pneumonia. The authors conclude that unexplained dyspnea after MMC therapy should raise the suspicion of PVOD and pulmonary toxicity. Of note, MMC may not be the only chemotherapeutic agent linked to PVOD, since an analysis has also shown that alkylating agents may represent a drug class that may be linked to PVOD development in susceptible individuals (122).

Although Perros et al. reported that a majority of MMC-induced PVOD patients died (4 of 7) in the French PH Registry despite PAH-specific therapies, there are conflicting case reports of MMC-induced PAH that improved with standard treatment. For example, Botros et al. report a case of a 60-yr-old woman diagnosed with anal carcinoma and then treated with MMC, 5-flurouracil, and local radiotherapy. Within 3 mo, diagnosis of PAH was established upon workup for symptoms of shortness of breath and ankle edema. PVOD was initially suspected; nonetheless, bosentan was added to diuretics and oral anticoagulation, and the patient recovered rapidly, no longer dependent on supplemental oxygen after 3 mo. Follow-up after 1 year confirmed that she had become symptom free while on the continued regimen that included bosentan therapy (13).

Selective Serotonin Reuptake Inhibitors

Selective serotonin reuptake inhibitors (SSRIs) are a recent addition to the list of PAH-associated drugs and toxins for their involvement in persistent pulmonary hypertension of the newborn (PPHN) (136). Multiple studies have been published with discordant results on the association of SSRIs with PPHN (5, 18, 19, 75, 147, 150, 151), but one population-based cohort study looking at over 1.6 million infants found that infants born to mothers who had taken SSRIs in late pregnancy were more than two times as likely to develop PPHN (79). Based on this large study, and the fact that serotonin signaling likely plays a role in pulmonary smooth muscle cell function (more on this in Aminorex, fenfluramine, and serotonin signaling), SSRIs are classified as having a “definite” association with PAH. However, PAH in adults exposed to SSRIs has not been thoroughly investigated.

Methamphetamine

Methamphetamine (METH) is a highly addictive and potent neurostimulant with systemic effects that can result in severe organ dysfunction and premature death (52). The molecular structure of methamphetamine is similar to that of aminorex fumarate and fenfluramine, stimulants known to cause D-PAH (2, 43) (Fig. 1). As of 2010, the worldwide prevalence of METH use was estimated at 14.3–52.5 million, second only to cannabis as the most common illicit substance of use. In California, METH use was associated with an incidence of 12–18 cases/10,000 hospital admissions between 2006 and 2013 (54). These epidemiological data stress the growing threat of METH to our society and the likelihood that practitioners will encounter patients afflicted with acute and chronic complications associated with METH use.

The link between METH and PAH has been reported by Schaiberger et al. and Chin et al. (21, 133); furthermore, PET studies have shown that intravenous d-[¹¹C]METH accumulates primarily in the lung and could serve as an active insult to the pulmonary circulation (143). Currently, METH use is considered a “likely” risk factor for the development of D-PAH (48), but it is evident that much more work needs to be done to clarify the clinical and pathological features of METH-induced PAH (METH-PAH).

Zamanian et al. recently reported our decade-long experience with METH-PAH and queried the Healthcare Cost and Utilization Project California database for admission related to METH abuse and PAH (156). METH-PAH was defined as PAH in the setting of significant METH exposure following documentation of more than three episodes of use per week for >3 mo. In our study population, most patients reported inhalation/smoking as the preferred administration route; the self-reported median METH exposure was 60 mo, which was similar across both genders. Most importantly, Kaplan-Meier analysis showed a 5- and 10-yr event-free survival of 47.2 and 25%, respectively, in METH-PAH vs. 64.5 and 45.7% in idiopathic pulmonary hypertension (IPAH) (50).

Using explanted lung tissue, we were able to document that lungs from METH-PAH patients demonstrated characteristic vascular changes similar to those of IPAH, including angiomatoid plexiform lesions with slit-like vascular channels within the artery (156). However, analysis of hemodynamic parameters showed that, compared with IPAH, METH-PAH is characterized by more severe pulmonary vascular disease at baseline, as evidenced by higher right atrial pressure, lower stroke volume index, and a more dilated and dysfunctional right ventricle. The severity of hemodynamic and right ventricle (RV) impairment in METH-PAH is worse than that described in dasatinib-PAH, further underscoring the harmful impact of this toxin on pulmonary vascular health.

HIV and Drug Abuse

Although individuals infected with HIV are at increased risk for developing PAH, incidence is more common in HIV-infected drug-abusing individuals. Multiple reports document increased risk of PAH development in HIV-infected individuals abusing illicit drugs such as cocaine, opioids, methamphetamine, and marijuana (22, 61, 69, 95, 102, 128, 154). Intravenous drug use (IVDU) accounts for one-third of all new HIV infection in the U.S. (8) and has been identified as the major risk factor of HIV infection in the individuals diagnosed with HIV-PAH. Various systematic reviews and reports on HIV-PH have reported IVDU to be a risk factor in 42–78% of HIV-infected patients diagnosed with PH (6, 72, 97, 99, 111). A report published by Quezada et al. (118) reported a 9.9% prevalence of PAH related to HIV among which 58.6% patients were intravenous drug users. In addition, another recent study (123) reported a PAH prevalence of 3% in a HIV-infected patient cohort that involved 51% intravenous drug users.

Enhanced pulmonary vascular remodeling in HIV-infected lung tissues from intravenous cocaine and/or opioids abusers (31) indicates that IVDU and HIV-1 potentially act in concert to cause a pulmonary arteriopathy. Studies demonstrate that the direct action of HIV proteins released by the infected lymphocytes and macrophages plays a major role in the development of HIV-PAH (71, 142). Corroborating these findings, it has been shown that cocaine exposure triggers severe pulmonary vascular remodeling and PH in a noninfectious HIV-transgenic (Tg) rat model expressing HIV-1 proteins (24, 90, 98). HIV-Tg rats exposed to cocaine (40 mg/kg) for 21 days had higher mean pulmonary arterial pressure and right ventricle systolic pressure (RVSP) compared with age-matched saline-treated HIV-Tg rats or wild-type rats given either cocaine or saline. Importantly, the combination of cocaine and viral proteins had a synergistic effect in augmenting pulmonary smooth muscle cell proliferation in the vasculature (24, 31). The lung tissues from simian immunodeficiency virus-infected rhesus macaques exposed chronically to morphine (opioid) also demonstrated a severe pulmonary vascular arteriopathy with obliteration of vessels resulting from enhanced proliferation of endothelial cells (137). Oxidative stress-mediated enhanced proliferation of endothelial cells was not only observed upon combined treatment of viral proteins with morphine but also with cocaine and methamphetamine when compared with either treatment alone (137).

Taken together, these findings suggest that illicit use of intravenous drugs by HIV patients acts as a second hit that modifies and potentiates pulmonary vascular remodeling, resulting in increased incidence and severity of HIV-PAH in HIV-infected drug abusers. Complicating the clinical scenario further, other variables often coexist in HIV-infected intravenous drug users such as opportunistic lung infections, hepatitis-related liver disease, or cigarette smoking, which may also contribute to the milieu conducive to pulmonary vascular injury.

Tobacco Smoke

The impact of tobacco smoking on the lung circulation remains a controversial topic in the field of D-PAH, yet the link between the development of PH secondary to small vessel loss from emphysema and destruction of lung parenchyma is not in doubt. Studies have shown that exposure to cigarette smoke extract can induce endothelial cell apoptosis (109) and also can trigger the production of antiendothelial antibodies in susceptible individuals (139). Animal studies have also shown that acute smoke exposure induces cell proliferation in pulmonary arterioles, ultimately leading to muscularization with chronic exposure and thus development of PAH (153). This may be in part the result of a significant increase in gene expression of endothelin and vascular endothelial growth factor, while exposed to cigarette smoke, with a disproportionately low expression of vasodilator-related proteins such as endothelial nitric oxide synthase (152). In 2009, Ferrer et al. investigated the contribution of cigarette smoking to PH in COPD at early disease stages (42). Nineteen male Hartley guinea pigs were exposed to daily cigarette smoke for either 3 or 6 mo and were compared with controls exposed to sham smoke during the same periods of time. When the animals were killed and pulmonary vessels were examined, it was found that endothelial dysfunction was seen at 3 mo of exposure and did not increase further after 6 mo of cigarette smoke exposure. At 3 mo, smoke-exposed samples showed poorly differentiated smooth muscle cells proliferating in small vessels. Furthermore, prolonged smoke exposure led to full muscularization of small pulmonary vessels, wall thickening, and increased contractility of the main pulmonary artery. Finally, lung expression of endothelial nitric oxide synthase was decreased in the group exposed to cigarette smoke.

In 2013, Keusch et al. performed an international, multicenter, observational, case-control study from 2008 to 2012 in which 472 patients with PAH and chronic thromboembolic pulmonary hypertension (CTEPH) were asked about their smoking exposure and habits (78). Included patients were from Switzerland, Germany, and Austria who met inclusion criteria of a mean pulmonary artery pressure ≥25 mmHg and pulmonary artery wedge pressure ≤15 mmHg. Those with concomitant chronic lung disease (WHO group III) and other confounding comorbidities (WHO group V) were excluded from the analysis; 233 of the included 472 patients (49%) were smokers and predominantly men (71%). The included cases were then compared against a group of healthy controls, and it was determined that significantly more PAH and CTEPH men were active smokers (71 vs. 57% of the healthy controls, P = 0.0007). When combining rates of active smoking and exposure to second-hand smoke, PAH and CTEPH patients had an overall higher active and passive tobacco smoke exposure compared with healthy controls, 68 vs. 51% in women (P < 0.0001) and 82 vs. 69% in men (P = 0.0002). Furthermore, active smokers among PAH and CTEPH women and men were significantly younger compared with nonsmokers (women: age 52 ± 13 vs. 58 ± 18 yr, P = 0.0051; men: age 57 ± 15 vs. 64 ± 17 yr, P = 0.0288). In all, the authors concluded that the increased prevalence of active and passive cigarette smoke exposure among the PAH and CTEPH population, along with a lower age at PH diagnosis in smokers, may indicate a pathogenic role of smoking. Finally, Trip et al. reported that a considerable number of patients in their Dutch cohort of IPAH patients were smokers that had a surprisingly low diffusion capacity for carbon dioxide, suggesting that in these patients there was a diminished pulmonary capillary bed (140).

Chemical Solvents

Occupational chemical exposure has been reported as a risk factor for PVOD and systemic sclerosis, a disease also associated with obstructive arterial and venous remodeling and PAH (33). A case-control study conducted at the National Reference Centre for Severe Pulmonary Hypertension in France showed that a significant occupational history of exposure was evident in 72.7% of PVOD patients vs. 27.7% of IPAH patients (116). Among PVOD patients, the most common occupational exposures included trichloroethylene (58.3%) with a median latency between first exposure and PVOD diagnosis of 48 yr and median exposure duration of 17 yr, suggesting that chronic and repeated exposure is probably necessary to induce PVOD. Compared with nonexposed patients, those with trichloroethylene exposure were older (54.8 ± 21.4 vs. 67.9 ± 8.1) and did not harbor any eukaryotic translation initiation factor 2-α kinase 4 (EIF2AK4) mutations, a risk factor for familial and sporadic PVOD. The authors suggest that individual susceptibility must be important and raises the possibility that polymorphisms in cytochrome P-450 2E1 (CYP2E1) could be linked to altered metabolism of trichloroethylene and could affect the susceptibility to trichloroethylene-induced toxicity in exposed patients (86).

Mechanisms of D-PAH Pathogenesis: Data From In Vivo and In Vitro Models

The last two decades have seen major advances in the understanding of the pathobiology of PAH. At present, severe PAH is thought to require the combination of genetic and environmental factors (i.e., the 2-hit hypothesis) that precipitate a phenotypic switch in the cellular milieu of the pulmonary vascular wall, resulting in increased pulmonary vasoconstriction, abnormal vascular remodeling, and right heart hypertrophy. D-PAH provides a unique window allowing the study of how environmental events such as exposure to certain drugs and/or toxins may precipitate such a cell phenotype switch in susceptible individuals. However, as we shall see in the following section, our understanding of D-PAH has been significantly impaired by the absence of animal models that recapitulate the clinical and pathological features of D-PAH upon exposure to known offending agents. For example, Byrne-Quinn and Grover administered aminorex or amphetamine to calves and assessed pulmonary hemodynamics after 4 wk of drug treatment; the animals did not develop PH (17). Mlczoch et al. treated pigs with fenfluramine alone or in combination with aminorex and found no evidence for PH or alterations of the pulmonary vascular tone (101).

PAH is a very complex and heterogeneous group of diseases in which multiple cell types, tissues, and even distant organs are thought to contribute to its development. The group of diseases is also characterized by several clinical features, not all of which may be present in each individual patient. Rodent models of PAH often do not recapitulate the prominent features of human pathobiology. Animal models of PAH use either chronic hypoxia or monocrotaline (MCT) to induce pulmonary endothelial injury, but a “second-hit” seems to be necessary to achieve an obliterative pulmonary vascular histopathology that resembles that observed in the human disease. For example, MCT combined with pneumonectomy or a shunt causes severe PAH with neointimal lesions and right ventricular hypertrophy, but MCT alone only caused moderate RVH and limited arterial muscularization (112, 149). More than 15 years ago, a new model of D-PAH had been developed. This model is built on the combination of a subcutaneous injection of the vascular endothelial growth factor (VEGF) receptor blocker Sugen 5614 (the first hit) with chronic hypoxia or pneumonectomy (the second hit) (1). In either case the rats reproducibly develop severe PAH because of a vasoobliterative microangiopathy, and they die from right heart failure. The drug Sugen 5416 causes apoptosis of lung vascular endothelial cells, hypothetically because the endothelial cells organ-specifically require intact VEGF signaling for their maintenance (3). Like the monocrotaline rat model, the Sugen 5416 models are models of drug-induced pulmonary endothelial cell injury. It is worth mentioning that cancer therapeutics that target VEGF are currently in use to treat various malignancies; despite one report on PH complicating severe cardiotoxicity with bevacizumab (i.e., Avastin) (20), no isolated cases of PAH have been associated with this drug.

As the list of offending agents continues to grow, it is imperative that we understand animal models better and develop additional models of D-PAH to gain a better understanding of the differences in metabolic pathways and detoxification responses compared with humans.

Aminorex, Fenfluramine, and Serotonin Signaling

The serotonin pathway has been identified by both experimental and clinical studies as a major contributor to disease severity and progression. Early studies that led to “The Serotonin Hypothesis” in PAH discovered a significant correlation between the consumption of a class of diet pills that augment serotonin signaling and the development of PAH in humans (2, 76). Treatment with one of these drugs, fenfluramine, has been shown to increase the level of circulating serotonin, or 5-hydroxytryramine (5-HT), by interacting with the 5-HT transporter (5-HTT) in platelets to cause the release of 5-HT (44, 94). PAH patients that have not been exposed to fenfluramine may also have elevated plasma 5-HT levels and increased expression of 5-HTT (37, 38, 65), placing serotonin signaling as a possible disease trigger. Serotonin can also signal through 5-HT receptors, including 5-HT_1b and 5-HT_2a. Studies in humans and rodents have identified 5-HT_1b to play a role in mediating vasoconstriction of small pulmonary arteries (77, 105). The contributions of 5-HTT and 5-HT_1b to the development of PH have been shown to be synergistic, since inhibiting both with chemical antagonists is effective in preventing 5-HT-induced PH (106). The effect of excessive serotonin signaling in the lung appears to occur through smooth muscle cells (SMCs), which respond to 5-HT with increased proliferation (85, 117), a hallmark of PAH. SMCs have also been shown to be resistant to apoptosis, with 5-HT signaling through 5-HTT and 5-HT_1b to activate the ERK pathway (88). Hypoxia is also found to increase circulating 5-HT levels in the airways and in cultured SMCs (36, 74), as well as to increase the expression of 5-HTT in SMCs located in the media of distal remodeled vessels of chronic hypoxic rats (36). SSRIs such as fluoxetine can competitively inhibit 5-HTT (4) and prevent the growth response of PA-SMC to 5-HT in vitro (36). Treatment with fluoxetine protects mice from chronic hypoxic-induced PH (93) and reverses MCT-induced PH in rats (55, 57). Moreover, fluoxetine has a suppressive effect on the chronic methamphetamine-induced pulmonary vascular remodeling in rats via inhibition of serotonin transporter 5-HTT (87). Although these studies support a therapeutic role for 5-HTT inhibition, a recent study by de Raaf and colleagues documented no difference in hemodynamics and lung angioproliferative changes in 5-HTT knockout and wild-type rats exposed to Sugen 5416 (29), questioning the role of serotonin in neointima formation and angioproliferative vascular remodeling.

The most important finding from animal studies investigating the roles of fenfluramine and dexfenfluramine is that these anorexigens indeed have an impact on lung homeostasis and vascular dynamics. Unfortunately, however, while in humans these drugs are strongly linked to disease onset, the results from fenfluramine and dexfenfluramine rodent studies have not produced a detailed understanding of the mechanisms of how these drugs contribute to the development of PAH. In fact, in one study, dexfenfluramine appeared to protect the rat lung vasculature against MCT-induced vascular disease (100). Similarly, Rochefort et al. found that the hypoxia-induced PH in rats was attenuated with continuous dexfenfluramine treatment (127). Other studies find that dexfenfluramine treatment does not affect the hypoxia-induced PH seen in rats after 4 wk, but discontinuation of treatment thereafter with continued hypoxia leads to severe PH with marked increases in 5-HTT expression and right ventricular hypertrophy (30, 35, 39). However, an earlier pair of studies from one laboratory found that in hypoxic dogs there was an increased vascular tone when the animals had been treated with dexfenfluramine. The conclusion of the authors was that the drug treatment was associated with a change in pulmonary vascular resistance that was unrelated to the vasoreactivity in response to hypoxia (107, 108). Chronic fenfluramine treatment in dogs was also found to lead to pulmonary vasoconstriction, but only in the presence of a second hit like endothelin-1 exposure (10). Collectively, these studies suggest that dexfenfluramine plays an active role in hypoxia-induced pulmonary vascular changes and suggest that the drug may strike a sensitive balance in how the lung tissue deals with oxygen availability.

Fenfluramine and dexfenfluramine are known to augment serotonin signaling by either increasing circulating 5-HT, 5-HTT expression, and/or 5-HT receptors with a notable effect in pulmonary smooth muscle cells. The Dempsie laboratory has described how anorexigens could have off target affects in the lung. This laboratory employed transgenic mice, either lacking tryptophan hydroxylase-1 or overexpressing 5-HTT, to ask how dexfenfluramine and the serotonin pathway cooperate to cause lung pathology. The results clearly showed that circulating 5-HT is necessary for dexfenfluramine treatment to cause vascular changes and increased pulmonary pressure, but the results also indicate that dexfenfluramine is protective against hypoxia-induced PH, and that this latter effect is likewise serotonin dependent (28). The serotonin pathway is also the subject of a study investigating whether METH generates lung disease in rats (145). Epidemiological studies place METH as a likely contributor to PAH, and the drug was also shown to cause pulmonary toxicity in mice (148). Wang et al. demonstrated that METH exposure in rats resulted in pulmonary toxicity with increased inflammation and arterial remodeling, and the authors argue that augmented serotonin signaling causes the observed phenotype. However, as is the case with other animal models of PAH, the rats do not present all of the clinical characteristics of PAH, since they do not develop increased pulmonary arterial pressure or right heart hypertrophy. Last, and of considerable clinical importance, severe PAH is not a hallmark of syndromes caused by serotonin-producing tumors.

BMPR2 Mutations and Serotonin

BMPR2 mutations are the most common mutations associated with heritable PAH, accounting for 80% of the cases; in addition, BMPR2 mutations can also be found in ~25% of cases of IPAH, supporting a key role in the pathogenesis (40). Interestingly, only ~20% of BMPR2 mutation carriers develop PAH, strongly suggesting that a “second hit,” perhaps also in the form of an environmental stressor, is required for disease development. One study has revealed that 9% of patients with fenfluramine-associated PAH are carriers of BMPR2 mutations (70). A pathological link between BMPR2 haploinsufficiency and serotonin exposure has been uncovered using BMPR2^+/− mice treated with a serotonin infusion. Compared with wild-type animals, infusion of serotonin resulted in increased pulmonary artery systolic pressure, right ventricular hypertrophy, and pulmonary artery remodeling in BMPR2^+/− mice, an effect that was further exaggerated under hypoxic conditions (an environmental stimulus known to trigger pulmonary vascular remodeling and vasoconstriction). Exposure to serotonin also resulted in greater suppression of BMPR2 signaling activity, as evidence by reduced phosphorylation (i.e., activation) of Smad1/5, the downstream signaling mediators of BMPR2, upon stimulation with the ligand BMP-2 (89). This suggests that serotonin may be a second hit required to uncover PAH in the face of BMPR2 haploinsufficiency (91).

Dasatinib

To date, there are no effective biomarkers nor any tailored therapies to treat individuals with dasatinib-induced PAH (103). Recently, Guignabert et al. used a combination of cell and animal models to elucidate the molecular mechanisms by which dasatinib provokes damage to the pulmonary circulation (56). Using clinically relevant doses (1 mg·kg⁻¹·day⁻¹) of dasatinib, the investigators pretreated 4-wk-old male Wistar rats before exposing them to MCT or hypoxia, stimuli known to trigger vascular remodeling and PH in rats. Compared with vehicle- or imatinib-pretreated controls, dasatinib-pretreated rats demonstrated significantly higher pulmonary arterial pressures, right ventricular hypertrophy, lower cardiac output, and vascular remodeling. Moreover, prominent infiltration of inflammatory cells (CD3⁺ lymphocytes, macrophages, and leukocytes) was observed around the remodeled vessels of the dasatinib-pretreated animals, suggesting a role for immune dysregulation in the pathogenesis of dasatinib-induced PAH. In addition, there was evidence for endothelial dysfunction as demonstrated by 1) an increase in the expression and release of markers of endothelial injury (intercellular adhesion molecule, vascular cell adhesion molecule-1, and E-selectin) in both animals and patients, 2) activation of the unfolded protein response (an endoplasmic reticulum stress response not seen with imatinib), and 3) reactive oxygen species (ROS)-dependent in vitro endothelial cell apoptosis that could be reduced by treatment with the antioxidant NAC. Taken together, these findings suggest that dasatinib could induce PAH in susceptible individuals by triggering ROS-dependent endothelial apoptosis, small vessel loss, and vascular remodeling.

MMC and PVOD

Following the description of their clinical cohort of PVOD patients with a history of MMC exposure, Perros et al. used rats to study the effect of MMC on the pulmonary circulation (116). A single dose of MMC (4 mg/kg single dose) was sufficient to produce PH with pulmonary arterial and venous remodeling that histologically resembled pulmonary capillary hemangiomatosis and PVOD. Interestingly, female rats were more prone to develop severe disease as evidenced by reduced survival and more severely altered hemodynamics. Mechanistically, it was proposed that MMC depletes levels of general control nonderepressible 2 (GCN2) protein in the lungs and mimics PVOD induced by mutations in the EIF2AK4 gene, which has been identified as the predisposing gene related to heritable PVOD. Of note, although there were no changes in BMPR2 levels, there was evidence of disruption of Smad1/5/8 signaling, suggesting a possible cross talk between BMPR2 and GCN2. In an effort to test therapeutic approaches, rats were given amifostine, a cytoprotective adjuvant commonly used to prevent nephrotoxicity by alkylating chemotherapy agents, simultaneously with MMC, resulting in significant improvement in survival, RV hypertrophy, and pulmonary hemodynamics compared with rats treated with MMC alone (116). This improvement was attributed to less muscularized distal microvessels and more low-resistance nonmuscularized distal microvessels, ultimately with a smaller percentage of occluded distal microvessels compared with rats treated with MMC alone. The authors conclude that simultaneous administration of amifostine may reduce the risk of developing PVOD.

Amphetamine and Methamphetamine

Amphetamine derivatives are metabolized by two major liver enzymes [CYP2D6 and carboxylesterase 1 (CES1)]. Although CYP2D6 is required for phase 1 metabolism of many drugs, CES1 is mainly involved in the detoxification of cocaine and heroin as well as FDA-approved stimulants (e.g., methylphenidate) (63, 157). Polymorphisms in these enzymes can contribute to chronic organ injury by affecting rate of metabolism (16, 26, 110, 138, 159) (81), but whether polymorphisms are linked to METH-PAH has not been explored.

To assess CES1 and CYP2D6 protein expression in pulmonary arteries, we performed immunofluorescence studies using lung tissue sections from two healthy donors and four METH-PAH patients obtained at the time of autopsy or transplant (113). Our studies demonstrated that CYP2D6 is expressed in the endothelium of small pulmonary microvessels; however, we found no difference in CYP2D6 expression between healthy donor and METH-PAH lungs (see The issue of drug metabolism by pulmonary cells regarding the role of CYP in pulmonary drug metabolism). In contrast, CES1 expression was found to be reduced or absent in remodeled vessels of all four METH-PAH samples compared with healthy patient samples (Fig. 2). To assess whether healthy human pulmonary microvascular endothelial cells (PMVECs) have the capacity to metabolize METH, we analyzed cell media and lysates of METH-treated PMVECs for metabolites using LC/MS. Analysis of the cell media did not identify significant amounts of METH or any known metabolites in either the treated or nontreated group, suggesting internalization and/or degradation of METH. However, analysis of cell lysates from METH-treated PMVECs demonstrated at least six hydroxyl isomers of METH produced via metabolic detoxification.

Fig. 2.

Carboxylesterase 1 (CES1) expression is reduced in vascular lesions of methamphetamine-induced pulmonary arterial hypertension (METH-PAH). A: representative immunofluorescence studies of lung sections stained for cytochrome P-450 (CYP) 2D6 (red) obtained from healthy donor and METH-PAH patients. No difference was seen among our 4 METH-PAH patients. B: representative immunofluorescence studies of lung sections stained for CES1 (red) obtained from healthy donor (top) and 4 METH-PAH patients. CD31 (green) stains for endothelial cells. Scale bar = 25 μm.

Further analysis of CES1 polymorphisms in the METH-PAH patient population led to the discovery of rs115629050. Found exclusively in METH-PAH samples, this missense single nucleotide variant is located within the active site of the enzyme, which could impair its detoxifying properties. Using a combination of genetic and mutation analysis, we found that reduction of CES1 via siRNA or introduction of a mutant form of CES1 carrying the single nucleotide variant led to greater in vitro PMVEC apoptosis in the setting of METH exposure. We next proceeded to explore the mechanism by which CES1 regulates PMVEC survival in the setting of METH exposure. Studies in neurons and in brain endothelial cells have shown that METH triggers production of ROS (25, 130) that results in activation of autophagy (12, 115, 121, 158). Autophagy is a process that can protect cells under stress until homeostasis is restored; in the setting of catastrophic injury, autophagy can facilitate apoptosis. Based on our findings, CES1 deficiency reduces the buffering capacity of autophagy in favor of apoptosis in response to METH exposure (Fig. 3).

Fig. 3.

Proposed model of the role of CES1 in METH-PAH.

Amphetamines and DNA Damage

Hypoxia is one of the major environmental triggers of PH and is known to increase the severity of hemodynamic impairment and vascular remodeling in the context of BMPR2 mutations, serotonin stimulation, and VEGF receptor blockade (32, 89). Interestingly, hypoxia can also facilitate genomic instability by suppressing DNA repair responses and increasing rates of mutagenesis. Based on this fact, Chen et al. postulated that the combination of hypoxia and amphetamine (or METH) exposure would increase the risk of PAH by promoting the accumulation of damaged DNA and more severe pulmonary vascular remodeling. Compared with healthy donor cells, PMVECs purified from lungs of METH-PAH patients displayed greater evidence of DNA damage as determined by comet assay and γH2AX foci. Furthermore, whereas treatment of cells with amphetamine alone did not induce DNA damage, costimulation with doxorubicin (a compound that induces DNA breaks) resulted in greater DNA damage and endothelial apoptosis. The investigators show that amphetamine exposure results in the destabilization of hypoxia-inducible factor (HIF)-1α and an increase in mitochondrial oxidation, caspase activation, and DNA damage. Complementary to the in vitro studies, the authors used a chronic hypoxia mouse model to test whether METH administration could lead to more severe PH and vascular remodeling compared with hypoxia alone. They found no difference in the level of RVSP or degree of RV hypertrophy, but treatment with METH resulted in more severe pulmonary vascular remodeling, as demonstrated by an increase in muscularized distal precapillary arteries. The authors speculate that exposure to amphetamine or METH can dysregulate the normal cellular response to hypoxia and facilitate the accumulation of pathological mutations in PMVECs raising the risk for PAH, the effect of which could be further aggravated in the presence of other PAH-associated mutations.

The Issue of Drug Metabolism by Pulmonary Cells

Because D-PAH is caused by drugs in genetically susceptible individuals and likely also genetically susceptible animals, it is worthwhile to review here some of the known aspects of lung drug metabolism and detoxification that is largely accomplished by two different lung cells (alveolar type II cells and lung microvascular endothelial cells). While there is presently no good reason to link PAH with normal or abnormal alveolar type II cell function, there is an overwhelming amount of data that describe metabolic functions of the PMVECs, which include quantitative uptake of drugs (for example, propranolol), mediators of inflammation such as serotonin, and precursor lipids like arachidonic acid, the angiotensin-converting enzyme-dependent conversion of angiotensin, and secretion of endothelin (9, 135). In addition to active uptake in PMVEC and secretion of factors, the PMVEC are also home to drug-metabolizing enzymes of the CYP450 group. It has been recognized that next to the liver the lung is the second important drug-metabolizing organ. Drugs can: 1) damage the lung endothelium directly, 2) induce the high expression of one or more CYP450 enzymes, which can generate cytotoxic drug metabolites, and 3) generate endothelial cell toxic or immunotoxic compounds. The drug-metabolizing CYPP450 isozymes are linked to the aryl hydrocarbon receptor (Ahr). The prototypical activating ligand of this receptor is tetrachlorodibenzodioxin; other ligands are kynurenine and components of cigarette smoke. Binding of the Ahr to the cytosolic nuclear transporter aryl hydrocarbon receptor nuclear translocator (ARNT, which happens to be the transcription factor HIF1-β) generates a transcriptionally active heterodimer that can bind to consensus regulatory sequences (xenobiotic response elements) located upstream in the promoter of target genes, which are CYP450 enzyme-encoding genes. For example, epidermal growth factor, TNF-α, and TGF-β are downstream targets of Ahr activation (60). Activation of Ahr also causes immunosuppression by modulating dendritic cell behavior (15). This Ahr-ARNT-CYP450 axis thus links drug metabolism with inflammation and immune responses (Fig. 4). It is of interest for the purpose of this review that dexfenfluramine induces CYP1B1 (27) and that Sugen 5416 is a powerful inducer of CYP1B1 (27-fold) and of CYP1A1 (260-fold). Table 1 lists a number of drugs and agents and their metabolism by various CYP450 isozymes or induction of CYP450 isozymes by the drugs; this list also includes trichloroethylene (82, 155), which is hypothesized to cause PVOD (see Chemical solvents).

Fig. 4.

The aryl hydrocarbon receptor (Ahr)-aryl hydrocarbon receptor nuclear translocator (ARNT)-CYP450 signaling pathway and its proposed role in drug-induced pulmonary arterial hypertension (D-PAH).

Table 1.

List of CYPs associated with different drugs known to produce D-PAH

Drugs	Cytochromes
Monocrotaline	Cyt3A4
Fenfluramine	Cyt1B1
Metamphetamine	Cyt 3A4, 2D6
Dasatinib	Cyt 3A4, 1A1,1B1
Interferon	Cyt 3A4, 1A2
Sugen 5416	Cyt 1A1, 1B1
Trichloroethylene	Cyt 4A10
Cigarette smoke components	Cyt 1A1, 2A6, 2E1, 3A

CYP, cytochrome P-450; D-PAH, drug-induced pulmonary arterial hypertension.

Finally, there are direct and indirect toxic effects of inhaled cigarette smoke components on lung cells, including lung vascular endothelial cells (141). Cigarette smoke components induce several CYP450 enzymes (114); they can alter estrogen metabolism and destroy lung capillaries following induction of endothelial cell apoptosis. Whether cigarette smoking facilitates the development of PAH in BMPR2 mutation carriers is unknown.

PREVENTING THE NEXT D-PAH EPIDEMIC

Physician Awareness

Because agents associated with drug-induced PH are newly identified, it becomes necessary to evaluate the current state of identification, communication, and education regarding culprit drugs and potential prevention of their toxic effects. From the point of view of the drug-prescribing physician, the importance of clinical screening and drug-taking history remains a crucial first step to identifying potential drugs that may inadvertently lead to PH. Thus, awareness of commonly reported associative agents, recognition of the pattern of injury (e.g., PAH vs. PVOD), and knowledge of diagnostic resources remain the current mainstay of initial suspicion for D-PAH.

Pharmacovigilance

In the U.S., the FDA is tasked with protecting public health by assuring the safety of human drug products. In 2006, to specifically address the process of identifying drug safety and toxicity, the FDA requested an evaluation by the Institute of Medicine. The resulting report identified three primary areas of focus to support a robust drug product safety program (drug safety science, operations and management, and safety communications). Within the FDA, the Center for Drug Evaluation and Research (CDER) is responsible for implementing these areas of focus. The CDER evaluates new drugs before marketing and performs postmarketing safety surveillance, including monitoring use of marketed drugs for unexpected health risks such as PH. The premarket review includes animal testing for toxicity across multiple species. Next, a phase 1 study is performed, with an emphasis in safety, and the most frequent side effects and drug pharmacokinetics are identified. Typically, this phase involves 20–80 subjects. After a phase 2 study emphasizes the desired effectiveness of the drug, a phase 3 study gathers more data about both safety and effectiveness, typically involving several hundred to 3,000 subjects. Finally, a new drug application is reviewed, and manufacturing facilities are scrutinized for safety purposes.

In 2008, the FDA launched its Safety-First Initiative, the mission of which was to provide equal attention to postmarketing drug safety as that given for premarketing review. Within the FDA, there are two major systems for postmarketing drug surveillance. First, the FDA Adverse Events Reporting System constitutes the “passive” system for identifying potential associations between a particular drug and toxicities. It comprises an electronic database of spontaneously submitted adverse events, also known as the MedWatch program, in which the public and drug sponsors or manufacturers voluntarily report cases of unintended side effects of marketed drugs. The intent is to identify serious, rare, or unexpected “safety signals” that may indicate new or unanticipated drug safety concerns. CDER then uses data-mining techniques to identify reports of greatest value. The second major system for postmarketing drug surveillance is the Sentinel System, the “active” system in which national integrated electronic databases are shared and integrated to monitor drug safety. The aim of this system is to proactively assess medical product safety in the real world, using collaboration between academic medical centers, healthcare systems, health insurers, and patient advocacy groups. Patient identifiers are removed, and large real-world data sets are then analyzed to draw associations between drugs and potential adverse events.

Limitations to the current FDA model for pharmacovigilance rest in the frequency and quality of reporting an adverse event. In the current passive system for adverse event reporting, MedWatch, 5% of reports annually are by the public, whereas 95% are by drug sponsors or manufacturers. This in turn brings to question the likelihood of reporting such events, dependent on the amount of time a product has been on the market or the degree of publicity around a particular adverse event. In addition, accuracy of reports is confounded by polypharmacy or comorbid patient conditions. Last, the CDER does not require a causal relationship to be proven before reporting, which may lead to inaccurate reporting of drug toxicity.

Drug Discovery

A major determinant of successful drug discovery is selection of agents with minimal toxicity. Monitoring for drug toxicity should be implemented early in the discovery process and maintained beyond drug approval, since some toxicities may not be evident in short-term studies. This is the focus of medicinal chemistry, a discipline devoted to improving the therapeutic and safety profiles of agents and whose participation in the drug development process can improve the quality of agents tested in clinical studies. Drug modifications that can improve the therapeutic index include chemical modifications to the molecular structure to improve solubility and biodistribution as well as designing novel delivery vehicles (e.g., nanocarriers, biogels) to target the drug to the pulmonary vasculature. In addition to the methods described above, there are numerous in silico tools (e.g., DEREK, TOPKAT, and MCASE) that can predict possible toxicity of a given candidate compound by analyzing its molecular structure and comparing it against a database of compounds with similar molecular structures and known toxicity profile. A major benefit of this approach is that structural analysis can help chemists modify the original molecule to improve target selectivity and pharmacological profile while reducing its toxicity.

Another approach to screening for drug toxicity is the use of patient-derived inducible pluripotent stem cells to generate differentiated vascular (endothelial, smooth muscle, fibroblasts) cells and cardiomyocytes. In addition to screening for efficacy of novel compounds, these cells provide a platform to screen the toxicity profile of multiple candidate compounds simultaneously with high-throughput technology in cells derived from either circulating blood or skin fibroblasts. Induced pluripotent stem cell-derived endothelial cells from PAH patients have become available and will likely help accelerate development of such platforms for the future (129).

Last, it is worth mentioning that animal models of PAH used for preclinical testing of compounds are rarely screened for evidence of possible systemic toxicity for the test compound. Animal models can be used to document and quantify in vivo toxicity of candidate compounds by searching for changes in clinical parameters (e.g., behavior, vital signs, EKG) and organ dysfunction through the study of explanted tissues or analysis of biomarkers in body fluids. One must also consider the toxicity of compounds such as Sugen 5416 and MCT, which are used routinely to induce PAH in rodents when it comes to assess potential treatment effects of drugs developed for the treatment of PAH.

Conclusions

In the last 50 years, numerous drugs and substances have been implicated in the development of PH (Table 2). One challenge in identifying drugs associated with PH is the relative rarity of PH and the frequent latency between starting a drug and the development of clinical symptoms. Another challenge is that animal models may not reproduce the most salient features of human forms of severe PAH, perhaps in part because the animals lack a required and human-specific genetic makeup. As a result, there continues to be a need for pharmacovigilance and the anticipation that there will be new drugs discovered that cause PAH in susceptible individuals.

Table 2.

Updated clinical classification of PAH

Definite	Possible	Likely	Unlikely
Aminorex	Cocaine	Amphetamines	Oral contraceptives
Fenfluramine	Phenylpropanolamine	l-Tryptophan	Estrogen
Dexfenfluramine	St. John's wort	Methamphetamines	Cigarette smoking
Toxic rapeseed oil	Chemotherapeutic agents	Dasatanib
Benfluorex	Interferon-α and -β
SSRIs	Amphetamine-like drugs

SSRI, selective serotonin reuptake inhibitors.

The drugs and agents highlighted here cover ~50 yr since the first epidemic of drug-induced PAH, and it is highly likely that more drugs will be identified as risk factors for PAH. In the past year, reports of PAH associated with other FDA-approved drugs, such as IFN and mitomycin, have been published, further illustrating the expanding role of prescription drugs as a cause of PH. As our understanding of the PAH pathobiology improves, we must pay attention to pathways and gene modifiers that participate in drug metabolism. Precision medicine tools such as gene sequencing, exosome analysis, and bioinformatics can serve in detecting genetic susceptibility factors that increase the risk of PAH in subsets of patients, but await validation in future studies.

In the meantime, physicians must remain vigilant of their patient population and establish close interactions between national drug regulatory agencies, national PH networks, and PAH patient associations to pursue leads and spread the word across the globe. We encourage clinicians to collect information regarding relevant drug exposure as part of the medical history and send out blood or urine for toxicology if there is suspicion of active stimulant use. Also, current guidelines recommend the use of FDA-approved vasodilators to treat patients with D-PAH as aggressively as those afflicted with other forms of WHO 1 PAH. Ultimately, prevention may prove to be the best strategy to deal with D-PAH, and it will help elevate awareness of this severe and often fatal disease.

GRANTS

V. A. de Jesus Perez is supported by National Institutes of Health (NIH) Grants R01-HL-134776 and R01-HL-134776-02, the Pulmonary Fibrosis Foundation, and the Pulmonary Hypertension Association. R. T. Zamanian is supported by NIH Grants NHLBI-HV-10-05, 1U01-HL-107393-01, PAR-09-185, and N01-HV-00242 and the Vera Moulton Wall Center. K. Yuan is supported by an American Heart Association Scientist Development Award and a Parker B. Francis Fellowship Award. N. K. Dhillon is supported by NIH Grants R01-DA-034542, R01-DA-042715 02, and 5 R01-HL-129875-02.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

M.E.O., K.Y., N.F.V., and V.A.d.J.P. prepared figures; M.E.O., K.Y., C.R., H.T., E.A.S., N.K.D., N.F.V., R.T.Z., and V.A.d.J.P. drafted manuscript; M.E.O., K.Y., C.R., H.T., E.A.S., N.K.D., N.F.V., R.T.Z., and V.A.d.J.P. edited and revised manuscript; M.E.O., K.Y., H.T., E.A.S., N.K.D., N.F.V., R.T.Z., and V.A.d.J.P. approved final version of manuscript.