Pulmonary Infection Within 100 Days After Transplantation Impaired Platelet Recovery in Patients with Hematologic Malignancies: A Propensity-Score-Matched Analysis

Background

Impaired platelet recovery occurs in 5–37% of patients who received allogeneic hematopoietic stem cell transplantation (alloHSCT) [1,2]. Several possible mechanisms may contribute to impaired platelet recovery, including impaired thrombopoiesis and increased platelet consumption [3]. Complications of alloHSCT, such as graft versus host disease (GvHD), hepatic venous sinus obstruction syndrome, thrombotic microangiopathy, and infections (bacterial, virus, and fungal), and their corresponding therapies are associated with impaired platelet recovery, even in the presence of full recovery of neutrophils [4–10].

Alterations in platelet counts during acute lung injury have been studies by several groups. Schneider and colleagues studied the fate of platelets in 15 patients with severe acute respiratory failure; among them, 10 patients developed thrombocytopenia (<100 000 platelets/microliters) [11]. In another study, Carvalho et al. studied platelet function in 13 acute respiratory failure patients admitted for intensive care, 6 acutely ill intensive care patients without evidence of acute lung injury, and 10 normal subjects [12,13], showing that patients with acute respiratory failure had quantitative and qualitative platelet defects that may contribute to thrombotic and hemorrhagic complications compared with those without acute respiratory failure.

Pulmonary infection occurs in 40–60% of alloHSCT recipients, and has been reported to be an important predictor of survival [14–16]. Previous studies have reported that the incidence rate of early-stage infections ranges from 15% [17] to 64.3% [14], and pulmonary infection is a common problem [15]. The reasons why pulmonary infection is associated with decreased survival of alloHSCT recipients mostly involve treatment-related complications [18], but few studies have focused on platelet recovery in patients with pulmonary complications during the alloHSCT procedure. The aim of the present study was to evaluate the effect of pulmonary infection on platelet recovery within the first 100 days after alloHSCT.

Material and Methods

PATIENTS:

From January 2011 to December 2018, 319 patients received alloHSCT at the Institute of Hematology, Changhai Hospital; 66 patients relapsed (21%), and 253 patients without relapse were included in the present study. Among these 253 patients, 62 patients (25%) developed pulmonary infection within 100 days. For each pulmonary infection case, a set of 2 patients was chosen as control from the remaining patients who were without pulmonary infection after matching for age (<20 y, 20–40 y, 41–60 y, or >60 y), disease and stage, time from diagnosis to transplantation, median CD34⁺ cells, and mononucleated cells transplanted. Fifty patients with pulmonary infection who had 2 matched controls were enrolled in a propensity score-matched study, and the remaining 12 patients who did not have matched controls were not included in propensity analysis. All procedures complied with the Helsinki Declaration standards and were approved by the Institutional Review Board of Changhai Hospital, Shanghai, China. The requirement for written informed consent was waived because the study used retrospective data from medical records, and there were no interventions performed in patients.

TRANSPLANT PROCEDURE:

The transplant procedure was described previously [19] Forty patients (40/50, 80%) received a busulfan and cyclophosphamide (BuCy)-based conditioning regimen that consisted of cytarabine (4 g/m2/d) intravenously on days −l10 to −l9; and Bu (3.2 mg/kg/d) intravenously on days −l8 to −l6; CTX (1.8 g/m2/d), intravenously on days −l5 to −4; Me-CCNU (250 mg/m2/d), orally once on day −3, and intravenously on days −5 to −2. Ten patients (10/50, 20%) were treated with an FBA conditioning regimen that included fludarabine, 30 mg/m2/day on days −10 to −6; Bu, 0.8 mg/kg every 6 h on days −5 to −3; and cytosine arabinoside, 1.5 g/m2/day on days −10 to −6). Anti-thymoglobulin (ATG) was used in patients who underwent alloHSCT from a human leukocyte antigen (HLA)-mismatched donor. The acute graft versus host disease (aGvHD) prophylaxis regimen was as previously reported [20].

INFECTION PROPHYLAXIS AND MONITORING:

Pulmonary high-resolution computed tomography (HRCT) was performed before alloHSCT. Patients were assigned to the horizontal laminar flow unit before conditioning. Sulfamethoxazole/trimethoprim was used to prevent prophylaxis Pneumocystis jiroveci. The preemptive treatment for CMV reactivation consisted of ganciclovir or valganciclovir. All patients received oral levofloxacin and fluconazole, unless they were receiving other medications for a previous infection. For secondary prevention, patients with history of invasive fungal disease (IFD) before alloHSCT received intravenous voriconazole or corresponding effective antifungal drugs. Within the first 3 months after alloHSCT, sulfamethoxazole/trimethoprim, fluconazole, and ganciclovir were administrated alternatively. CMV DNA, Epstein-Barr virus (EBV), procalcitonin (PCT), C-reactive protein (CRP), 1, 3-β-D-glucan assay (G assay), and galactomannan test (GM assay) were measured once a week until 3 months after alloHSCT.

DIAGNOSIS OF PULMONARY INFECTIONS AND TREATMENT:

Samples were collected from the sputum, blood, and bronchoalveolar lavage (BAL) fluid if possible. Chest HRCT was performed when patients were highly suspected to have a pulmonary infection. PCT and CRP, as well as G and GM assays, were measured. Cultivation or next-generation sequencing of BAL fluid or lung biopsy were performed to confirm pathogenic organisms.

Once pneumonia was diagnosed, empirical broad-spectrum antibiotics were instituted if there was suspected bacterial pneumonia until a specific pathogen was identified, and targeted therapy was initiated accordingly. Active IFD was defined as possible, probable, or proven according to the EORTC criteria [21]. The specific antifungal treatment was administrated at the discretion of the managing physician. In addition, we used PSI (Pneumonia Severity Index) scores to weigh the prognosis of pulmonary infection [22].

DEFINITIONS OF CLINICAL OUTCOME:

Platelet count ≥20×109/L for 7 consecutive days without transfusion was defined as platelet engraftment (primary platelet recovery). Secondary failure of platelet recovery (SFPR) was diagnosed in patients who had previously fulfilled the criteria for trilineage recovery after alloHSCT, with platelet count <20×109/L or ≥20×109/L with transfusion for more than 7 consecutive days [23,24]. Good graft function (GGF) was defined as persistent successful engraftment (ANC > 0.5×109/L for 3 consecutive days, PLT counts >20×109/L for 7 consecutive days and hemoglobin levels > 70 g/L without transfusion support) beyond 28 days after HSCT [25]. Prolonged thrombocytopenia (PT) was defined as the engraftment of all peripheral blood cell lines other than a platelet count ≤80×109/L for more than 90 days after alloHSCT [9,23]. Impaired platelet recovery was included in PT and SFPR. CMV and EBV viremia were defined as >1000 viral copies/ml plasma by PCR. Diffuse pulmonary infection was defined as diffuse lesions in ≥2 pulmonary lobes. Scattered pulmonary infection was defined as lung consolidation of 1 lung segmental lobe, or single and multiple lung nodules based on CT scan.

Chimerism analyses were performed by DNA fingerprinting for single-nucleotide polymorphism in bone marrow samples as introduced in our previous publication [26] (sensitivity >0.01% recipient signals).

STATISTICAL ANALYSIS:

Greedy-type 1: 2 matching technology without replacement was performed to match patients with pulmonary infection and patients without (Figure 1). Matching of the logit of the propensity-score began with high accuracy (0.00001), and then gradually decreased to minimum precision accuracy (0.1). Univariate and multivariate logistic regression were used to examine the risk factors associated with platelet recovery. The considered fixed risk factors included the following: age at transplantation (age <40 y vs. age ≥40 y), gender, disease type, HLA disparity (matched vs. mismatched), donor–recipient gender matching (others vs. female–male), ABO compatibility (matched vs. mismatched), donor–recipient relationship (related or unrelated), conditioning regimen (without ATG vs. with ATG, without CTX vs. with CTX, without Flu vs. with Flu), aGvHD, CMV/EBV viremia, pulmonary infection, upper respiratory infection, gastrointestinal infection, and blood stream infection. Survival data were analyzed with Kaplan-Meier method and compared with log-rank test. Overall survival (OS) was defined as the date from alloHSCT to death. Treatment-related mortality (TRM) curves were constructed in the competing risks framework, considering death without relapse after alloHSCT as a competing event. Variables are presented as frequency and percentage and compared using the chi-square test or Wilcoxon rank sum test. A P value less than 0.05 was defined as statistically significant. SAS 9.4 (SAS Institute, Inc., Cary, NC, USA) and R3.4.2 were used in statistics analysis.

Results

PATIENT CHARACTERISTICS:

Among the 253 evaluable patients, 54% (136/253) were male. The median age was 36 years (range, 11–72 years) and the median follow-up was 36 months. Within 100 days after alloHSCT, 25% (62/253) of patients had a pulmonary infection, and 34% (87/253) of patients received unrelated donor alloHSCT. The mean time for neutrophil and platelet engraftment were 13.34±2.66 days and 15.19±8.32 days, respectively. The incidence of aGvHD for the whole cohort was 23% (59/253). The probabilities of OS and TRM at 3 years after alloHSCT were 79.5±2.7% and 20.53±0.07%, respectively. After propensity-score-matched analysis, 50 patients with pulmonary infection and 100 patients without included in further analysis. The clinical characteristics of the 50 patients with pulmonary infection are shown in Table 1.

PROPENSITY-SCORE-MATCHED ANALYSIS:

Propensity-score-matched analysis included 150 patients (50 patients with pulmonary infection and 100 patients without). The median follow-up was 28 months in cohorts with pulmonary infection and 38 months in those without. There was higher incidence of pulmonary infection in patients who received grafts from HLA-mismatched donors compared with HLA-matched donors [53.8% (21/39) vs. 26.1% (29/111), P=0.002. CMV and EBV viremia were more frequent in pulmonary infection patients. The difference in the incidences of grade II–IV aGvHD between the groups was not significant (28% vs. 22%, P=0.417; Table 2).

Pulmonary infection occurred at a median time of +32 days [range: −10 to +97 days]. The incidences of pulmonary infection were 10% (5/50) during conditioning, 36% (18/50) in the first 30 days after alloHSCT, 40% (20/50) at 60 days, and 12% (6/50) at 90 days after alloHSCT. Fungal pneumonia (30/50, 60%) was more common than bacterial pneumonia (14/50, 28%) and viral pneumonia (1/50, 2%). Polymicrobial pneumonia was diagnosed in 4 patients. In 30 cases of fungal pneumonia, 5 were proven and 12 were probable (Table 1). There were 12 patients diagnosed pneumonia with specific pathogenic organisms through sputum cultivation (5/12, 42%) or invasive diagnostic procedure, including microbial culture and next-generation sequencing of BAL fluid (5/12, 42%), and lung biopsy (2/12, 16%).

The mean times for neutrophil engraftment were 14.55±4.18 days and 12.93±1.93 days (P=0.012), and 19.29±13.96 days and 13.94±4.12 days (P=0.012) for platelet engraftment for patients with and without pulmonary infection, respectively. Among patients with pulmonary infection, 2 patients failed to engraft, and the other 48 patients achieved full donor engraftment as measured by SNP-PCR, in which 33% (16/48) patients achieved GGF, 21% (10/48) patients with SFPR, and 46% (22/48) patients with PT.

The incidence of PT in patients with pulmonary infection was 44% (22/50), and 9% (9/100) in the corresponding matched group (P<0.001). Megakaryocyte counts in patients with pulmonary infection were lower than that in patients without pulmonary infection [median number: 36 (range, 0 to 450) vs. 44 (range, 0 to 1000)/slice (1.5×3 cm²), P=0.010]. Seventeen patients (17/50, 34%) developed pulmonary infection before platelet engraftment. The platelet engraftment was delayed in these patients than in patients without pulmonary infection [mean time: 28.33±21.91 days vs. 13.94±4.12 days, P=0.024]. Of them, 10 patients (10/17, 59%) developed PT, and the percentage was markedly higher than in patients without pulmonary infection (9/100, 9.00%; P<0.001). The difference in time for platelet engraftment between patients with diffuse (n=9) and scattered pulmonary infection (n=39) was not significant. (20.11±12.35 days vs. 19.10±14.45 days, P=0.848)

Thirty-three patients developed pulmonary infection after primary platelet recovery, of whom 24% (8/33) patients developed SFPR. The platelet counts at 90 days after alloHSCT in patients with pulmonary infection were lower than in patients without pulmonary infection. [median: 55×10⁹/L (range, 2–188×10⁹/L) vs. 128×10⁹/L (range, 8–308×10⁹/L), P<0.001]

To determine the specificity of the impact of pulmonary infection on platelet recovery, we enrolled patients with upper respiratory tract infection, gastrointestinal infection, and blood stream infection during alloHSCT as non-pulmonary infection controls. There was a weak relationship between infections in other sites and impaired platelet recovery (infection vs. non-infection: upper respiratory tract infection: 13.61±2.59 days vs. 13.83±4.45, P=0.731; gastrointestinal infection: 13.71±4.04 days vs. 13.83±4.45, P=0.919; blood stream infection: 17.21±15.02 days vs. 13.83±4.45, P=0.343).

UNIVARIATE AND MULTIVARIATE ANALYSIS:

Univariable logistic regression analysis contained patient characteristics such as age (<40 years vs. ≥40 years), sex (female vs. male), conditioning regimen, donor characteristics, transfused cells, and infections. Univariable logistic regression showed that impaired platelet recovery was strongly associated with HLA mismatch (OR 2.638, 95% CI: 1.450–4.798, P=0.001), CMV viremia (OR: 3.306, 95% CI: 1.726–6.331, P< 0.001), grade II–IV aGvHD (OR: 2.025, 95% CI: 1.079–3.801, P=0.028), and pulmonary infection (OR 6.103, 95% CI: 3.237–11.505, P<0.001, Table 3). These 4 factors were then enrolled in the multivariable analysis. Multivariable logistic regression showed that pulmonary infection was a significant independent risk factor for impaired platelet recovery (OR: 5.335, 95% CI: 2.735–10.407, P<0.001, Table 4).

SURVIVAL OUTCOMES:

The probability of OS at 3 years after alloHSCT was 71.5±6.4% for patients with pulmonary infection and 80.2±4.2% for patients without pulmonary infection (P=0.132). The patients were further grouped into 2 subgroups depending on infection severity. Grade III–IV infection conferred a worse OS at 3 years (54.2±10.2% vs. 88.1±6.4%, P=0.014) compared to patients with grade I–II infection (Figure 2). The probability of OS in patients with pulmonary infection in the first month (n=18) was 66.7±11.1%, and 76.9±8.3% for patients (n=26) who developed pulmonary infection during the second and third months. (P=0.410).

The probability of OS at 3 years after alloHSCT was 55.2±7.3% in the group with impaired platelet recovery and 87.9±3.6% in patients with GGF (P<0.001). Patients with PT or SFPR had inferior survival compared with patients with GGF (3-year OS: PT: 69.9±8.4%, P=0.005, Figure 3A; SFPR: 29.4±11.1%, P<0.001, Figure 3B). TRM was increased in patients with impaired platelet recovery (impaired platelet recovery: 44.78±0.55% vs. GGF: 12.12±0.13%, P<0.001). The probabilities of TRM at 3 years were 30.13±0.74% in patients with PT and 70.59±1.39% in patients with SFPR (Figure 3C, 3D).

Discussion

To the best of our knowledge, this study is the first on platelet recovery and pulmonary infection within 100 days after alloHSCT. We performed propensity score matching to balance baseline patient characteristics. The incidence of PT in patients with pulmonary infection were 44% (22/50) and 9% (9/100) in the corresponding matched group (P<0.001).

Despite the advances in prophylaxis and therapy of infections, pulmonary infections remain a major cause of death in more than 40% of alloHSCT recipients [27–29]. In the present retrospective study, we identified 62 cases of pulmonary infection (62/253, 25%) during the transplant procedure, in which pneumonia was diagnosed based on clinical symptoms, microbial culture, biochemical assay, invasive diagnostic procedures, and imaging results. Fungal infection was the leading etiology for pneumonia within 100 days after transplantation. Patients received levofloxacin prophylaxis in our study, which was reported in previous studies to lower the proportion of early bacterial infection after transplantation [14,30]. Although CMV viremia was documented in 33 out of 150 patients, only 1 patient developed CMV-related pneumonia. Dynamic monitoring of the CMV viral load in plasma and improved preemptive anti-viral therapy contribute to the lower incidence of CMV pneumonia. As expected, the incidences of CMV and EBV viremia were still higher in patients with pulmonary infection than in patients without (Table 2). Pulmonary infection delayed the immune reconstitution, which facilitated the reactivation of viruses. Myelosuppression caused by CMV infection itself, and ganciclovir-induced cytopenia, were more difficult to manipulate, both of which contributed to impaired platelet recovery, and then shorter survival. Of the 10 patients who developed SFPR, 3 died of aGvHD and 3 died of respiratory failure. Four patients who had CMV viremia received ganciclovir preemptive therapy and recovered from SFPR. PT and SFPR were associated with shorter OS and higher TRM, consistent with a previous study [24]. The underlying association between aGvHD and impaired platelet recovery has been extensively studied by others and by our group [10,23,31,32]. We have suggested that bone marrow may be a potential aGvHD target through immune-mediated cytopenias. In the present study, although the distribution of aGvHD was balanced in the 2 groups, aGvHD still had an impact on platelet recovery in the whole cohort in univariate analysis.

In 36 surviving patients with pulmonary infection, only 15 (42%) achieved GGF, which shows that pulmonary infection may have a close relationship with platelet recovery. Pulmonary infection was a risk factor for impaired platelet recovery [OR 5.335, 95% CI (2.735–10.407), P<0.001]. Of note, infections in other sites did not have any overt impacts on platelet recovery, which suggests that platelet recovery is related to a direct impact on pulmonary disruption rather than a systemic inflammatory state.

Infection, iron overload, aGvHD, CMV infection, the use of ganciclovir or valganciclovir, and in vivo T cell depletion are significantly associated with increased risk of SFPR [10,23,24,31,33]. Thrombopoiesis, which occurs within a specialized bone marrow microenvironment, is a complex biological process that is initiated with the differentiation of hematopoietic stem cells (HSCs) to megakaryocytic progenitors and eventually results in the maturation of megakaryocytes to produce functional platelets [18,34]. A recent study showed that the lungs in mice dynamically released platelets, representing almost 50% of total platelets production [35]. In a state of thrombocytopenia and relative stem cell shortage in the bone marrow, those progenitors can repopulate the bone marrow, and contribute to multiple hematopoietic lineages [36]. Therefore, lung injury is associated with reduced number of megakaryocytes from the lungs, and this may result in thrombocytopenia.

Our study had several limitations. First, the analysis was retrospective. Second, small size is a major limitation in our study, which made it difficult to comprehensively analyze the potential variables affecting platelet recovery. However, our propensity score-matched model showed that pulmonary infection has a powerful and specific effect on platelet recovery. Multi-center clinical trials on this topic are needed.

Conclusions

The incidence of impaired platelet recovery in patients with pulmonary infection following alloHSCT was relatively high. Pulmonary infection during alloHSCT is related to an increased risk of impaired platelet recovery (SFPR and PT) and is related to worse OS and higher TRM. Further research should concentrate on the underlying mechanism of pulmonary infection and impaired platelet recovery, which could help improve the prognosis of alloHSCT recipients.