Pulmonary complications occur in approximately 25% and 27% of pediatric and adult HSCT recipients, respectively, and are a leading cause of transplant-related mortality (5). Overall, about 9% of HSCT recipients require invasive ventilation and 10% require non-invasive ventilation (4–11). However, it is likely that pulmonary complications are underreported in both adult and pediatric patients. Roychowdhury et al. (12) reviewed autopsies and bronchoalveolar lavage (BAL) slides of patients who died after HSCT and determined that pulmonary complications occurred in 40 (80%) of the 50 cases and were a major contributor of transplant-related mortality in 74% (37 of 50). Furthermore, Inaba et al. (13) evaluated the incidence of abnormal pulmonary function tests in 89 transplant survivors and demonstrated that abnormal pulmonary function testing was seen in 40.4% of baseline testing and 64% of post-HSCT testing.

Time of onset from cell infusion is helpful when evaluating causes of pulmonary complications: pre-engraftment (0–30 days), post-engraftment (30–100 days), and late phase (over 100 days) (14). The pre-engraftment period is marked by neutropenia, mucositis, indwelling lines, and acute graft-versus-host disease (GVHD) (14–17). Post-engraftment phase is marked by impaired cellular immunity and acute GVHD. The late phase is marked by impaired humoral/cellular immunity and chronic GVHD (14, 17, 18). In all phases, severe GVHD (II–III), acute or chronic, increases relative risk of pulmonary complications. One study notes an increase in relative risk of 2 with a 95% CI of 1.1–3.7 (4, 19).

Pulmonary complications after HSCT can be subclassified as infectious vs. non-infectious ( Table 1 ), related to the way impairment of immune function differs over time after HSCT. Pulmonary complications caused by infectious vs. non-infectious etiologies have incidence ranging from 13.9%–54.8% to 10.2%–39.7%, respectively.

Bacterial pneumonia is the most prevalent type of infectious complication in all phases, with an incidence up to 45% (15, 21, 23, 24). In the pre-engraftment phase, the major causative agents include Gram-negative organisms (such as Pseudomonas aeruginosa) due to the poor mucosal barrier, acute GVHD, central lines, and protracted neutropenia (14, 15, 22, 25). In the late phase, infectious etiologies more commonly feature encapsulated organisms and are associated with chronic GVHD. This is likely due to continuation of immunosuppressive medications (14, 15, 25, 26).

Nocardia infection has a low cumulative annual incidence. One study noted a cumulative annual incidence of 1.75% throughout the course of treatment, predominantly occurring in the late phase (15, 27–29). Nocardial infection is more common in patients with allogeneic HSCT, those with a history of acute GVHD, and those actively being treated for chronic GVHD at the time of diagnosis (27, 28). Additional risk factors include other concurrent infections, in particular, CMV infection (27, 28). Nocardia tends to be disseminated at diagnosis but commonly has a pulmonary locus (30). There is some evidence that those who receive pentamidine prophylaxis for Pneumocystis jirovecii also have increased risk, but this is not consistently demonstrated throughout the literature (28, 30).

Mycobacterium infections including both tuberculosis and non-tuberculous subtypes occur at low incidence worldwide (0.1%–5.5%) and are more prevalent in those who have received an allogeneic HSCT. In the United States, incidence has been reported from 0.0014% to 3% (31–33). Infection typically occurs in the late phase, and while Mycobacterium infections can be disseminated, infection is predominantly in the lungs (31, 32). Tuberculosis infection is associated with older age and chronic GVHD (34). Most recommendations are for conservative management, with treatment if the patient has a tuberculosis exposure even with negative skin testing. Currently, there is no evidence for prophylaxis as incidence remains low (32). Studies have demonstrated that it is likely safe to treat after Day +100, and treatment consists of isoniazid, rifampin, ethambutol, and pyrazinamide for 6–9 months (33).

Overall, the reported incidence of fungal infections ranges from 4% to 34%, occurring most commonly in allogeneic HSCT patients and during the post-engraftment and late phases. Mortality can be up to 33.3% (3, 4, 17, 23). The most common fungal pulmonary complications in HSCT patients are invasive aspergillosis, followed by invasive candidiasis, then mucormycoses (17, 35). In one large multicenter study, incidence rates of each fungal infection were reported as 43%, 28%, and 8% respectively (35). There is an increase in fungal infections with protracted/continued neutropenia (60 days or longer) and concurrent GVHD (35).

Invasive aspergillosis has a reported incidence in autologous HSCT of 1% to 5%, most frequently diagnosed in post-engraftment and late phases (15, 35). Invasive aspergillosis cases have continued to decline with the integration of granulocyte colony-stimulating factor and azole prophylaxis in treatment (14, 25, 36). Aspergillosis is diagnosed using a combination of radiologic and serologic testing. Serum galactomannan and aspergillosis serum PCR testing can be sent for diagnosis, but the most sensitive/specific test is the BAL galactomannan (14, 25, 37). The accuracy of testing has been shown to be related to neutrophil count and underlying condition (38). The current recommended treatment is with antifungals such as voriconazole or amphotericin B (25, 37).

Overall, invasive candidiasis infections have been decreasing in incidence, in particular, Candida albicans. There has been an increase in the rate of Candida glabrata and Candida krusei, likely secondary to antifungal prophylaxis (15, 39); one study demonstrated that 70% of infections were attributed to that non-albican species when analyzing autologous HSCT recipients with Candida infections (35). Candida infections span the entire course of transplant, peaking in the post-engraftment phase before the first 120 days (35). Diagnosis is made through fungal cultures from both serum and BAL. Initial treatment of choice for Candida species is echinocandins (such as micafungin) or voriconazole, with further modification based on sensitivities of cultures (40).

Mucormycosis infections are also increasing in incidence with the use of azole antifungal prophylaxis (25, 41). One study showed an incidence of 8% with infections typically occurring in the late phase (35). CT scans can be helpful for diagnosis and can show reversed halo signs (an area of central ground-glass necrosis surrounded by a ring of consolidation); however, for diagnosis, BAL with fungal cultures are needed. Treatment is with amphotericin B and accompanied by surgical resection if without significant morbidity/mortality (14, 25, 41).

Of note, P. jirovecii pneumonia is a rare complication after HSCT, with an incidence of 0.63% in allogeneic HSCT and 0.28% in autologous HSCT recipients (42). In one study, when a patient was diagnosed, they were 6.87 times more likely to die when compared to their matched controls (42). Patients are at an increased risk if they have GVHD and/or poor immune reconstitution (42). As this complication is a rare cause of pulmonary complications, it will not be discussed in significant detail in this review.

Viral pneumonia in all HSCT patients ranges between 4% and 21.9%, with a greater incidence in allogeneic HSCT patients. The most commonly reported viral infections are Cytomegalovirus (CMV) and adenovirus (17, 21, 23). In a study evaluating the changes in rates of infection over time, CMV disease fell from 8% to 5% from 1993–1997 compared to 2003–2007 (3). Mortality greatly varies between types of viral infection, viral quantities, and autologous vs. allogeneic HSCT. Respiratory syncytial virus (RSV), influenza, and parainfluenza are other common etiologies of viral pneumonia that are seen in HSCT patients (14).

CMV becomes a major cause of pneumonia starting at 3 weeks posttransplant and continues into the late phase. There have been improvements in infection rates through careful selection of donors, careful serology monitoring, and early intervention. The greatest predictor of CMV infection is the serology status of the recipient (14, 15, 23, 25, 43, 44). Autologous HSCT patients have a reported incidence of 1%–9% (15, 44). Again, acute GVHD and allogeneic HSCT patients have an increased risk of CMV infection (14, 43). Diagnosis requires radiologic and positive CMV PCR from BAL or viral cultures (14, 15). Mortality still approached 31% in one study focusing specifically on autologous HSCT recipients. While letermovir can be used for prophylaxis, it has not been used to treat active infections (45, 46). Treatment of CMV infection includes gancyclovir and foscarnet with/without CMV immunoglobulins (14, 15, 44). Resistance testing can be performed if continued breakthrough viremia or increasing viral count while on treatment. Some centers treat more persistent or severe infections with virus-specific T cells (targeted therapy) (14).

While a common community-acquired entity, RSV infection in HSCT patients (as well as rhinovirus, parainfluenza, and human metapneumovirus) has been shown in studies to progress to pneumonia between 35% and 58% of the time (15, 47, 48). These infections occur equally in autologous and allogeneic HSCT (15). Additional risk factors for developing a significant pneumonia from these common viruses include severe lymphopenia, T cell-depleting or myeloablative conditioning, and acute GVHD (47, 48). RSV can be particularly severe in HSCT patients, leading to additional complications such as acute respiratory distress syndrome or even diffuse alveolar hemorrhage (DAH) that results in invasive respiratory support (4, 49, 50). One study reported that up to 10% of their patients developed acute lung injury from RSV (50). Chemoprophylaxis with palivizumab in high-risk children (RSV outbreak or young infants) has been shown to have some benefit in prevention of RSV infection (47). Rapid RSV PCR is diagnostic, and in severe cases, aerosolized ribavirin and immunoglobulins can be considered, though there are little data to support an improvement in mortality (14, 25, 49).

Other viral infections that should not be excluded from evaluation are human herpesvirus 6 (HHV6), herpes simplex virus (HSV), and adenovirus. Going forward, additional studies will be needed on the novel coronavirus disease 2019 (COVID-19) virus in HSCT patients.

PERD typically occurs within the first 5 days of engraftment (pre-engraftment phase) and is accompanied by established clinical criteria, with pulmonary edema seen on imaging (25). Clinical criteria include fever, rash, hepatic or renal dysfunction, weight gain, hypoxemia, and transient encephalopathy (15, 53). The etiology is not completely understood but thought to be secondary to pro-inflammatory cytokines (54). PERD does have a higher incidence in autologous HSCT patients, with studies showing an incidence between 2.5% and 20% (5, 14, 54, 55). Overall incidence seemed to increase over time, which is thought to be secondary to the introduction of granulocyte colony-stimulating factor during HSCT (56). Typically, there is a good response to steroids (14, 15, 53).

IPS usually presents with fever, acute respiratory distress, and alveolar damage that has an unknown underlying etiology (not caused by infection or end organ damage) (8, 51, 52). Currently, the best evaluation of etiology has come from murine models that suggest conditioning regimens including lung irradiation, cyclophosphamide, busulfan, or previous treatments with carmustine (BCNU), etoposide, bleomycin, and cisplatin all increased the risk of epithelial injury. This leads to activation of pulmonary macrophages and alloreactive T lymphocytes. Implicated cytokines include interleukin-6, interleukin-8, and in particular tumor necrosis factor (TNF)-alpha (52, 56, 57). Imaging is non-specific with multilobular infiltrates. BAL is typically performed to rule out an underlying infectious process (52, 58). Onset usually is in the later portion of pre-engraftment to the beginning of the post-engraftment phase (21, 51). There is a higher incidence in allogeneic HSCT with a mean up to 10% in allogeneic HSCT patients and 5.8% in autologous HSCT patients (14, 25, 51, 58). The onset for autologous HSCT is typically later in the mid-to-late post-engraftment phases (21, 58). Risk factors specifically for developing IPS include chest irradiation, older age, being female, or solid tumor diagnosis (58). Other risk factors that seem to increase the incidence of IPS include high-dose cyclophosphamide and adding busulfan to the conditioning regimen (52). Mortality is significant from 60%–80% in allogeneic transplants, with few recent studies specifically looking at mortality in autologous transplant patients (58). The patients who develop this complication have the highest rate of mortality once intubated (approaching 74%) (59).

Currently, first-line treatment with corticosteroids and supportive care is recommended; however, studies are mixed whether there is improvement in respiratory support or outcome (52, 59). Etanercept (TNF alpha binding protein), has been shown to reduce pulmonary vascular endothelial cell apoptosis; one study evaluating the combination of steroids and Etanercept demonstrated improvement in mortality, increasing D +28 survival to 73% (15, 57, 60). Thus, a randomized, double-blind, placebo-controlled trial was initiated comparing etanercept to placebo; all patients received methylprednisolone. This study was unable to produce significant results in mortality/outcome; however, the trial’s results are difficult to interpret as it was halted due to poor study enrollment (25, 61). This demonstrates again an area that necessitates further investigation to find alternative treatment options and evaluate ways to improve outcomes. Treatment with agents such as etanercept should only occur after confirmation that there is not an underlying infectious etiology. IPS additionally encompasses subcategories including DAH and COP, which will be discussed later.

DAH is a subcategory of IPS that is defined by hemorrhagic alveolitis. In pediatric and adult allogeneic HSCT, incidence of DAH ranges from 5% to 12% with a median onset of 19 days as compared to autologous HSCT recipients where DAH incidence ranges from 2.1% to 12%, with a median time of onset of 12 days (5, 58, 62, 63). Definitive diagnosis requires a BAL sample with at least 20% of hemosiderin-laden macrophages, blood in 30% of alveolar surfaces, with increasingly bloody samples (15, 58, 64). In one study, DAH was associated with engraftment, an age over 40, solid malignancies, high fevers, severe mucositis, and/or with renal insufficiency (65). Currently, treatment recommendations include high-dose steroids and supportive care (5, 62). Small studies have been performed in patients with DAH (both transplant induced and not) that have evaluated inhaled/nebulized tranexamic acid ± recombinant activated factor VII with reported success; however, further investigation with a larger sample size is required before integrating this into standard of care (63, 66, 67). Despite early intervention and supportive care, mortality historically has been up to 80%–100% in allogeneic HSCT patients. Newer studies suggest that mortality for autologous HSCT patients is closer to 28% and 70% for allogeneic patients (21, 58, 62, 64, 67). If DAH occurs in the first 30 days, mortality is significantly lower (64).

Drug toxicity has a varying degree of severity—from mild dyspnea to respiratory failure requiring mechanical ventilation. Incidence ranges from 22% to 49% in autologous HSCT patients with a mean onset at Day +45 (post-engraftment phase); however, there is a wide range from 21 to 149 days (52, 58 68). Drug toxicity is seen more frequently in patients who have had regimens including BCNU, etoposide, cyclophosphamide, bleomycin, and cisplatin (52, 58, 68, 69). Radiologically, CT scans are not specific but can demonstrate patchy ground-glass opacities, sometimes with septal thickening. Biopsy can show hypersensitivity reaction with eosinophilic pneumonia or, if performed later, a thickening of the interstitium with early fibrosis (51, 52). The treatment of choice is corticosteroids, though little data are available on the overall mortality of this complication (58).

COP tends to be a more subacute process with fever, dyspnea, and cough (51). Incidence typically ranges from 0.9% to 10% and can occur in both allogeneic and autologous HSCT patients (14, 15, 51). For allogeneic patients, there is a well-described association with chronic GVHD, thus suggesting an immune-mediated response (51). For autologous HSCT patients, the proposed mechanism is secondary to underlying infection or drug toxicity (51). COP usually occurs within the first 100 days and occurs more frequently in allogeneic HSCT patients. On chest CT, patchy consolidations with elongated distribution and ground-glass opacities can be seen (14, 51, 70). Diagnosis is confirmed via biopsy that shows patchy plugs of granulation tissues that fill lumens of distal airways (alveoli) with chronic interstitial inflammation and no prominent bronchiolar damage (15, 52). Corticosteroids remain the treatment of choice with a slow taper, usually with significant improvement. In one study, 78%–80% of patients demonstrated a good response (14, 15, 52, 71). However, there are no studies looking at what duration of steroid therapy is most appropriate. Additionally, COP tends to have a widely varied reported relapse rate of 9%–58%, which is likely secondary to no standardization of treatment across institutions (52, 70).

Pulmonary hypertension (PHTN) is characterized by increased pulmonary vascular resistance and elevated right ventricular pressure (72). The incidence of PHTN is not well defined but has been estimated between 15% and 28% with a mortality up to 55%–86% (73, 74). This complication most commonly occurs in the late phase with a reported median of Day +70; however, the range is wide from Day 0 to +365 (72). The underlying etiology in HSCT is not well understood; the proposed mechanism is endothelial injury in both pre/post-pulmonary capillary vasculature, leading to smooth muscle proliferation, fibroblast infiltration, and ultimately hypertrophy of the vasculature (72, 75). This damage can be instigated by other underlying disorders such as thrombotic microangiopathy and atypical hemolytic uremic syndrome (72, 73). Chemotherapy agents that have been implicated include mitomycin, bleomycin, cisplatin, vincristine, cyclophosphamide, and BCNU (76). Additional risk factors include high-dose preparative chemotherapy and radiation prior to transplantation. Diagnosis is usually obtained via electrocardiogram and echocardiogram, but sometimes more invasive measures are implemented, such as transesophageal echocardiology or cardiac catheterization (72, 75). Biopsy can be obtained to confirm the diagnosis and shows widespread fibrous proliferation in the pulmonary venules and small veins; however, this is not typically performed due to the morbidity and mortality associated with the procedure (76). Treatment includes supportive care, oxygen therapy, inhaled nitric oxide, calcium channel blockers, and phosphodiesterase-5 inhibitors (72). For long-term damage caused by PHTN, high-dose steroids can be used, though there are no studies evaluating the effectiveness of this treatment (76, 77). There are limited studies evaluating this complication, and further investigation is warranted.

BO exclusively occurs in allogeneic HSCT patients and is one of the most common causes of late-phase complications (from Day +90 to 2 years) (15, 25, 51). Incidence ranges from 2% to 48%; this wide range is likely secondary to the inconsistencies in diagnostic criteria between studies (51). Associated features include acute GVHD, older age, non-related donor, total body irradiation, peripheral stem cell source, and busulfan-based conditioning regimen (14, 25, 51, 52). BO is a histologic diagnosis that can be made after lung biopsy, which shows constrictive bronchiolitis and submucosal bronchiolar fibrosis (15, 25, 51, 78). However, lung biopsies are not without significant morbidity and mortality. Thus, clinical criteria have been established through the National Institutes of Health Consensus Development Project to describe bronchiolitis obliterans syndrome (BOS). This requires four features to meet diagnostic criteria. First, FEV1/VC <0.7 (or 5th percentile of predicted). Second, FEV1 <75% of predicted with ≥10% decline over less than 2 years. FEV1 should not correct >75% with albuterol. Third, no identified infection. Fourth, evidence of air trapping/bronchiectasis on high-resolution chest CT or residual volume >120% of predicted value (79). While this is a useful outline, studies have shown that not all patients with histologic BO meet the established criteria of BOS. Treatment can include both inhaled and systemic steroids with other immunosuppressive medications, though very few studies demonstrated significant clinical benefit (51). For patients whose symptoms are steroid refractory, treatment can include Janus-associated kinase 1/2 inhibitors such as ruxolitinib (80, 81). Studies have been published reporting 59%–68% overall response rate with a tolerable safety profile (80, 82). Most common toxicities included reactivation thrombocytopenia and anemia; some studies have reported increased CMV reactivation, though this is not consistently demonstrated across all studies (80–83). However, these studies evaluated those 12 years of age or older; studies evaluating effectiveness/dosing in children younger than 12 and dosing are ongoing (81, 83). Other agents that are currently under investigation include belumosudil and ibrutinib, though further investigation is needed for use in pediatrics (84, 85). Despite aggressive treatment, most patients have acute flares with mortality rate being between 12% and 27% at 5 years due to secondary infection or respiratory complications (8, 51, 78).

Other entities that can cause pulmonary complications that are not discussed here include acute GVHD, pulmonary alveolar proteinosis, pulmonary cytolytic thrombi, and chronic GVHD.