CT-guided Core-Needle Biopsy of the Lung Is Safe and More Effective than Fine-Needle Aspiration Biopsy in Patients with Hematologic Malignancies

Abstract

Purpose

To evaluate the safety and diagnostic yield of CT-guided core-needle biopsy (CNB) versus fine-needle aspiration biopsy (FNAB) of lung nodules and masses in patients with hematologic malignancies (HMs).

Materials and Methods

With institutional review board approval, 166 patients were retrospectively reviewed between 2007 and 2017 who were diagnosed with leukemia, lymphoma, or myelodysplastic syndromes (with or without hematopoietic stem cell transplant) and who underwent CT-guided FNAB and/or CNB of the lung. Patient medical records, pathologic reports, and interventional biopsy reports were reviewed.

Results

In the study period, 166 patients underwent percutaneous CT-guided lung biopsy; 36% (60 of 166) of the procedures included CNB (CNB + FNAB and CNB only), whereas 64% (106 of 166) were FNAB only. In the CNB group, FNAB was also performed for 92% (55 of 60) of the patients before CNB; 13% (eight of 60) of patients in the CNB group were nondiagnostic versus 45% (48 of 106) of FNAB only (P < .0001). There was no statistically significant difference in the pulmonary complication rates, with 1.7% of CNB and 1.9% of FNAB only requiring chest tube placement (P = .7), 5% of CNB and 2.8% of FNAB only developing hemoptysis (P = .4), and 5% of CNB and 2% of FNAB only developing hemothorax (P = .3). A change in clinical management was observed in 51% of patients with diagnostic biopsies compared with 21% of patients with nondiagnostic biopsies (P = .0002).

Conclusion

CT-guided CNB is an effective technique for performing lung biopsy in patients with HMs with higher diagnostic yield compared with FNAB, and a higher, although not a statistically significant, increased risk of bleeding complications and pneumothorax.

© RSNA, 2019

See also the commentary by Elicker in this issue.

Summary

CT-guided core-needle biopsy of the lung in patients with hematologic malignancies is a safe and effective technique without an increase in the risk of biopsy-related bleeding complications and pneumothorax compared with fine-needle aspiration biopsy.

Key Points

• ■ Percutaneous lung biopsy complication rates in patients with hematologic malignancies (HMs) are similar to those in the general population.

• ■ Core-needle biopsy has higher diagnostic yield than does fine-needle aspiration biopsy for lung biopsies in patients with HMs, with higher, although not a statistically significant, increased risk of complications.

Introduction

Patients with hematologic malignancies (HMs) experience a variety of pulmonary complications including bacterial infections, fungal infections, recurrent or secondary malignancies, organizing pneumonia, drug-induced pneumonitis, and granulomatous diseases. It is estimated that as many as 13%–60% of patients with HMs develop some form of pulmonary complication during the course of the disease (1,2). Patients with HMs who undergo hematopoietic stem cell transplant (HSCT) are at further increased risk of infectious (3) and noninfectious complications (4,5). The common noninfectious complications after HSCT include pulmonary edema, pulmonary hemorrhage, pulmonary leukostasis, engraftment syndrome, and therapy-related complications (4,5). Although new and more effective treatment options for HMs have improved overall survival (6), recurrent disease, subsequent malignancies, and pulmonary and cardiac diseases remain as some of the leading causes of mortality (7).

Diagnosis of respiratory complications is challenging given the complexity of these cases and the overlapping clinical presentations of infectious and noninfectious entities. Bronchoscopy with bronchoalveolar lavage, transbronchial biopsy, open lung biopsy, and percutaneous CT-guided biopsy are commonly performed to obtain tissue and cultures for diagnostic work-up. CT-guided biopsies can be performed with fine-needle aspiration biopsy (FNAB), core-needle biopsy (CNB), or a combination of both. Effective treatment depends on the biopsy results, including antimicrobial therapy targeted at the specific fungus or bacteria, steroids for noninfectious processes, or chemotherapy targeted at a malignancy. These considerations make CT-guided biopsies a vital component in the management of HMs. The complications associated with CT-guided FNAB or CNB include pneumothorax (which may require the placement of a chest tube), hemothorax, hemoptysis, air embolism, and death. In general, the occurrence rates of major complications are relatively low for CT-guided procedures, ranging between 4% and 6% (8).

Patients with HMs are considered to be at higher risk for invasive procedures given thrombocytopenia, platelet dysfunction, immunosuppression, anemia, and transfusion dependence. Although de Bazelaire et al (9) reported an overall complication rate of less than 5% for lung biopsies, many other studies have shown an increased risk of complications for CNB versus FNAB, with reported complication rates of 20%–40% for CNB and 15%–24% for FNAB (8,10). Hence, operators may be reluctant to use CNB in this population. However, in general, CNB has higher diagnostic yield than FNAB (10–14). In particular, the success of FNAB depends on obtaining a sufficient number of viable cells (14) and is influenced by the expertise of the operator and the cytopathologist. CNB, on the contrary, has the advantage of obtaining larger amounts of tissue. In addition, CNB was considered relatively operator independent in one of the studies, although this study included only a small cohort of patients (15).

To our knowledge, prior studies of CT-guided lung biopsies in patients with HMs have not evaluated the diagnostic yield and outcomes of FNAB versus CNB. The aim of this study was to evaluate the diagnostic yield and complication rate of FNAB versus CNB in patients with HMs. In particular, we hypothesized that CNB would have a higher diagnostic yield than FNAB for CT-guided lung biopsy in this patient population.

Materials and Methods

Patient Selection

Institutional review board approval was obtained, and informed consent was waived for our Health Insurance Portability and Accountability Act–compliant retrospective observational cohort study. We searched the electronic medical records of two large tertiary care academic hospitals for patients with diagnostic coding of leukemia, lymphoma, or myelodysplastic syndromes with or without HSCT who had undergone a CT-guided lung biopsy (CNB and/or FNAB) from 2007 to 2017. Patients who had undergone nonlung biopsies (for example, mediastinal, lymph node, chest wall, or pleural biopsy) were excluded. A total of 166 patients were included in the study group with 101 (61%) men and 65 (39%) women. A total of 55 patients who had undergone nonlung biopsies were excluded. Median age of the included patients was 59 years with a range of 25–87 years. Charts were manually reviewed by one author (G.V.W.), a thoracic radiology fellow with 1 year of training in thoracic radiology to confirm the underlying diagnosis and biopsy procedure of the patients.

Record Review

Electronic records were reviewed to evaluate whether the FNAB or CNB technique was used to obtain tissue. Complications were classified as pneumothorax with or without placement of a chest tube, hemothorax, hemoptysis, and death. The biopsy results were classified as malignancy, bacterial infection, fungal infection, organizing pneumonia, nonspecific inflammation, nondiagnostic, or other. Nondiagnostic was defined by no growth from cultures and a nondiagnostic histopathologic report. Nonspecific inflammation classification included all pathologic reports citing the presence of an inflammation, for example, granulomatous inflammation, scar with chronic inflammation, or acute inflammation with necroinflammatory debris. The “other” category included pathologic reports that were not citing a malignancy, an infection, or an inflammation and included terms such as atypical epithelioid cells, atypical pneumocytes, pulmonary parenchyma with dense fibrosis, and focal alveolar hyperplasia. Records were reviewed for the presence of a change in patient management after biopsy. Management change was defined by change in medications (antibiotics and/or antifungal therapy), chemotherapy, or surgery. For inpatients, management changes were considered change in management either during the same admission to the hospital or during follow-up clinic visits based on the biopsy results. For outpatients, management change was considered any treatment change during the follow-up visits in the clinics based on the biopsy results.

For comparison of the complications and biopsy result rates, a control group was selected from the institution’s thoracic biopsy procedure log. This group consisted of 336 consecutive patients who underwent CT-guided lung biopsy at Brigham & Women’s Hospital from July 2017 through June 2018. Pathologic diagnosis and complications were recorded in the log per institution policy.

Procedure Protocol and Policies

The standard protocol for the patients scheduled for lung biopsy was to review laboratory values for the patients, including hemoglobin level (>8 g/dL), platelet count (≥50 K/μL), and international normalized ratio ([≤1.5]), performed within 30 days before the procedure, consistent with the Society of Interventional Radiology Guidelines (16). Partial thromboplastin time was not included in routine preprocedural testing. Furthermore, all patients were screened for a history of unexplained bleeding disorders, anticoagulation medications, or bleeding risk factors (sepsis, dialysis, chemotherapy, and disseminated intravascular coagulation). Anticoagulation medications were withheld after consultation with the referring physician regarding the risks and benefits. Our institution’s policy is to discontinue aspirin 5 days before biopsy, to discontinue prophylactic enoxaparin 12 hours and therapeutic enoxaparin 24 hours before biopsy, and to discontinue therapeutic heparin 6 hours before biopsy.

Procedures were performed with a dedicated multislice CT scanner by using CT fluoroscopy with the use of a sterile technique. The choice of local anesthesia only versus local anesthesia with conscious sedation was at the discretion of the operator. CNB was performed with the coaxial needle technique by using either 17-gauge to 18-gauge needles (introducer-core-needle combination) or 19-gauge to 20-gauge needles (introducer-core-needle combination). FNAB was performed by using either 20-gauge or 22-gauge needles, either with coaxial or single-needle technique (depending on operator preference). The choice of FNAB, CNB, or both for the biopsy technique to be used was decided by the operator. Onsite cytopathology was used for biopsies, including for FNAB samples, per standard institutional practice.

Following completion of the procedure, outpatients were observed for at least 2 hours in the recovery room. Two radiographs were obtained, one immediately after the procedure and one 2 hours after the procedure. If a new or enlarging pneumothorax was observed on the second chest radiograph, then a third radiograph would be obtained at up to 4 hours after the procedure. Inpatients would undergo the first radiograph in the recovery room and, after a brief period of observation, would go back to their hospital room for further observation and then undergo a second radiograph at approximately 2 hours later.

Data Entry and Statistical Analysis

Data were entered in REDCap (17) and analyzed with JMP Pro version 14 (SAS software; SAS Institute, Cary, NC). Data were tabulated to evaluate the raw difference in diagnostic yield (nondiagnostic rate) and complication rates, comparing the FNAB group with CNB group by using a Fisher exact test. A similar comparison was made by using the raw diagnostic yield and complication rates for the HM and control groups, stratified by FNAB or CNB. In addition, a multivariable logistic regression model analysis was performed for bleeding complications (hemoptysis and hemothorax) by including CNB, platelet count, and HSCT as variables.

Tests for statistical significance of categorical variables were performed with a one-tailed Fisher exact test as noted above, with α set at .05. A power calculation was performed with the program G*Power (version 3.1; University of Dusseldorf, Germany). Given the observed rate of 1% hemoptysis in the 291 control CNB group patients, with α of .05, we had 80% power of detecting an increased rate of hemoptysis, if it is at least 8.5% in the 60 patients with HM who underwent CNB.

We performed propensity score matching by using SPSS (version 23; IBM, Armonk, NY). We included platelet count and HSCT as potential independent variables influencing the risk of major bleeding after biopsy. Using the CNB group (n = 58) as cases, the propensity score algorithm selected 58 matched FNAB-only controls (using a threshold of 0.05). A McNemar test was performed to analyze the difference in bleeding complications between the matched CNB and FNAB-only cohorts.

We also performed a Fisher exact test to evaluate the effect of diagnosis and prior therapy on the rate of management change after biopsy. A separate multivariable logistic regression model incorporating the features of diagnostic biopsy and prior antifungal therapy was used for calculation of the impact of biopsies on change in management.

Results

Patient Clinical Characteristics and Diagnoses

A total of 166 patients underwent percutaneous CT-guided biopsy (Table 1). The platelet counts at the time of biopsy ranged between 38 K/μL and 658 K/μL with a median of 145 K/μL. One patient in the study who underwent FNAB only had a count less than the threshold of 50 K/μL (ie, 38 K/μL) at the discretion of the interventional radiologist.

Table 1:

Clinical Characteristics and Biopsy Results

The underlying patient diagnoses included acute myeloid leukemia (42%, 69 of 166), non-Hodgkin lymphomas (21%, 35 of 166) including 10 diffuse large B-cell lymphomas (6%, 10 of 166), myelodysplastic syndrome (9%, 15 of 166), chronic lymphocytic leukemia (8%, 14 of 166), acute lymphocytic leukemia (7%, 12 of 166), Hodgkin lymphoma (4%, seven of 166), and other diagnoses (8%, 14 of 166); 40% of patients had undergone HSCT before the lung biopsy.

Among 166 patients who underwent CT-guided biopsy, 36% (60 of 166) of patients had undergone CNB (CNB + FNAB and CNB only), whereas the remaining 64% (106 of 166) of patients had undergone FNAB only. In the CNB group, FNAB was also performed for 92% (55 of 60) of the patients before performing CNB. Thus, in total, there were five CNB-only patients, 55 FNAB and CNB patients, and 106 FNAB-only patients. A total of 56 (34%, 56 of 166) CT-guided procedures were nondiagnostic. In the diagnostic group, of 110 (66%, 110 of 166) patients, 33 (30%, 33 of 110) had malignancies, 32 (29%, 32 of 110) had fungal infections, 10 (9%, 10 of 110) had bacterial infections, 17 (15%, 17 of 110) had nonspecific inflammation, 12 (11%, 12 of 110) had organizing pneumonia, and six (5%, six of 110) had other causes. Of the 42 infections where specific agents were identified, there were 17 (40%, 17 of 42) cases of Aspergillus; 10 (24%, 10 of 42) cases of Mucor; 10 (24%, 10 of 42) cases of bacterial infections including three of Nocardia, two of mycobacteria, two of Pseudomonas, and three of other bacteria; and five (12%, five of 42) cases of infections caused by other fungi.

Within the control group of unselected lung biopsies at our institution, 45 (13%, 45 of 336) were FNAB only, and CNB samples were obtained from 291 (87%, 291 of 336) patients. FNAB samples were also obtained from 87 (30%, 87 of 291) patients before performing CNB.

Diagnostic Yield

In the study group, eight (13%, eight of 60) CNBs were nondiagnostic versus 48 (45%, 48 of 106) of FNAB only (P < .0001) (Table 2). In patients who underwent both CNB and FNAB, the nondiagnostic rate was 15% (eight of 55). In the control group, 4% (12 of 291) of CNBs were nondiagnostic versus 31% (14 of 45) of the FNAB-only group (P = .01).

Table 2:

Diagnostic Yield and Complications of CNB (CNB + FNAB and CNB Only) versus FNAB-Only Group

Complications

There was no statistically significant difference in complication rates between FNAB-only and CNB groups (CNB + FNAB and CNB only) in the HM cohort, with 2% (one of 60) of CNB patients requiring chest tube placement versus 2% (two of 106) of FNAB-only patients (P = .7) (Table 2). The overall pneumothorax rate was 18% (11 of 60) in the CNB group (CNB + FNAB and CNB only) versus 25% (26 of 106) in the FNAB-only group (P = .2) and 18% (10 of 55) for the CNB + FNAB group. Composite bleeding complications were defined as the presence of either hemoptysis or hemothorax. Hemoptysis was seen in three (5%, three of 60) patients of the CNB group (CNB + FNAB and CNB only), three (3%, three of 106) patients of the FNAB-only group (P = .4), and three (5%, three of 55) patients of the CNB + FNAB group. Hemothorax was seen in three (5%, three of 60) patients of the CNB group (CNB + FNAB and CNB only), two (2%, two of 106) patients of the FNAB-only group (P = .3), and three (5%, three of 55) patients of the CNB + FNAB group. There was one death in the FNAB and CNB group and none in the FNAB-only group. A propensity score–matched cohort of 58 FNAB-only samples compared with 58 in the CNB group (CNB + FNAB and CNB only) yielded a composite bleeding complication rate of 5.2% (three of 58) for FNAB-only compared with a rate of 8.6% (five of 58) for the CNB group (P = .727).

The comparison between the diagnostic yield and complication rates between the study group of patients with HM and control group of unselected patients is provided for the CNB, FNAB-only, and CNB + FNAB groups in Tables 3–5. In the control group, none of the patients who underwent FNAB only had hemothorax or hemoptysis. In the CNB control group, complication rates were 2% (five of 291) for hemothorax and 1% (three of 291) for hemoptysis (Table 3). Rates of patients with pneumothorax requiring chest tube placement were 4% (two of 45) for FNAB-only controls versus 3% (eight of 291) for CNB controls. There was no statistically significant difference in these rates compared with those of the HM cohort.

Table 3:

Diagnostic Yield and Complications of CNB Group (CNB + FNAB and CNB Only) Biopsies (Patients with Hematologic Malignancy versus Controls)

Table 5:

Diagnostic Yield and Complications of CNB + FNAB Group (Patients with Hematologic Malignancy versus Controls)

Table 4:

Diagnostic Yield and Complications of FNAB-Only Group (Patients with Hematologic Malignancy versus Controls)

We evaluated the effect of platelet count at the time of biopsy on complication rate between patients with counts less than 75 K/μL and more than 75 K/μL at the time of biopsy. There was no statistically significant difference in the risk of chest tube insertion for pneumothorax (0% vs 2%; P = .6) in patients with counts less than 75 K/μL. There was a higher rate of hemoptysis with lower platelet count that did not reach statistical significance (11% vs 2%; P = .06). Similarly, the rate of hemothorax was greater in the low platelet count group, but this also did not reach statistical significance (7% vs 2%; P = .06). Of note, the median platelet count of patients who underwent CNB (CNB + FNAB and CNB only) was slightly higher than that for FNAB only (151 K/μL vs 130 K/μL) patients, but this was not statistically significant (P = .3).

The only death in the study occurred from massive hemoptysis in a patient with myelodysplastic syndrome with HSCT who underwent FNAB + CNB in the prone position with a 19-gauge to 20-gauge (introducer-core-needle combination) needle of a right lower lobe nodule. The patient had a platelet count of 51 K/μL at the time of biopsy and developed massive hemoptysis immediately following completion of the procedure. Posttransplant lymphoproliferative disorder was diagnosed on conducting histopathology.

Composite bleeding complications were seen in eight (12%, eight of 67) HSCT patients compared with two (2%, two of 99) non-HSCT patients (P = .01), indicating that HSCT is a risk factor for bleeding complications (Table 6). Of note, within the bleeding complications, there was a substantially higher rate of hemothorax (7%, five of 67) in patients with HSCT versus (0%, 0 of 99) non-HSCT patients. A multivariable logistic regression model analysis for bleeding complications (hemoptysis and hemothorax) including CNB, platelet count, and HSCT as variables showed that the only statistically significant variable in the model was HSCT with an odds ratio of 6.32 (P = .02). The odds ratio for CNB and platelet count was 2.13 (P = .27) and 0.99 (P = .16), respectively.

Table 6:

Complication Rates in Patients with or without Hematopoietic Stem Cell Transplant

When complication rates were analyzed based on specific diagnoses (Table 7), there was a statistically significant difference in hemoptysis rates with 2% (two of 98) of patients with leukemia, 2% (one of 43) of patients with lymphoma, 7% (one of 15) of patients with myelodysplastic syndrome, and 20% (two of 10) in the other group developing hemoptysis (P = .048). There was no statistically significant difference in pneumothorax rates (P = .93), chest tube rates (P = .52), or hemothorax rates (P = .16) for patients based on a specific disease. In addition, no significant difference was observed in rates of nondiagnostic biopsies (P = .49).

Table 7:

Diagnostic and Complication Rates from Underlying Diagnosis

Management Change and Medications before Biopsy

Among the 110 patients who underwent a diagnostic biopsy, 56 (51%, 56 of 110) experienced an alteration in clinical management following the biopsy versus 12 (21%, 12 of 56) patients who underwent a nondiagnostic biopsy (P = .0002). Among the 81 patients receiving antifungal medication before biopsy, 43 (53%, 43 of 81) had a management change versus 25 (36%, 25 of 70) who were not receiving antifungal medication (P = .02). Among the 42 patients diagnosed with infections (bacterial or fungal), 29 (69%, 29 of 42) experienced management change versus 39 (31%, 39 of 124) for other diagnoses including nondiagnostic biopsies (P < .0001). A multivariable logistic regression model incorporating these features (diagnosis and antifungal therapy) showed that only the diagnosis was statistically significant (for infectious diagnosis, P = .0006; for nondiagnostic biopsies, P < .0001). There was no statistically significant difference in the rate of management changes for CNB versus FNAB only, 24 (40%, 24 of 60) versus 44 (41%, 44 of 106), respectively (P = .64).

Among the 32 patients with fungal infections, 26 (81%, 26 of 32) were undergoing antifungal therapy before undergoing biopsy. The percentage of nondiagnostic cases was slightly higher in patients receiving antifungal therapy (38% vs 27%), but this did not meet statistical significance (P = .1).

There was a statistically significant difference in management change based on specific disease with 50 (51%, 50 of 98) patients with underlying leukemia, 12 (28%, 12 of 43) with lymphoma, four (27%, four of 15) with myelodysplastic syndrome, and three (30%, three of 10) with other diagnoses getting a management change (P = .046).

Discussion

In summary, we evaluated 166 patients with HMs who underwent CT-guided lung biopsies. We found that CNBs (CNB with or without FNAB) had a greater diagnostic yield than FNAB only (87% vs 55%). Although there was a higher rate of complications in the CNB group than in the FNAB-only group, statistical significance was not met. However, there was one biopsy-related death in the CNB + FNAB group, with none in the FNAB-only group. We found that patients who underwent diagnostic biopsies were more likely to undergo a change in clinical management than those who underwent nondiagnostic biopsies. We also found that patients with a platelet count less than 75 K/μL had a nonsignificant increase in bleeding complications, and patients with a history of HSCT had a statistically significant increase in bleeding complications.

The difference between the yield of CT-guided FNAB versus CNB in HMs has, to our knowledge, not yet been explored. Our study found that 87% of CNBs were diagnostic compared with 55% of FNAB only. This is most likely related to the larger amount of tissue obtained by using CNB relative to FNAB. In the literature, the overall diagnostic rates for CT-guided biopsies range from 35% to more than 90% in various studies across a spectrum of diseases (18–21). CNB has been shown to more readily provide an adequate amount of tissue for histology and culture (11,12,15). Of note, one previous article showed a similarly increased yield of CNB versus FNAB of the lung in patients with solid organ transplants (22).

Infections comprise a large fraction of the pulmonary process in patients with HMs, as opposed to routine healthy outpatients with lung nodules in whom malignancy is by far the most common diagnosis. The two most common diagnoses in our cohort were infection (35%) (Fig 1) and malignancy (30%) (Fig 2). These findings are comparable with reported incidence in patients with HM from prior studies, which range from 29% to 34% for infections and 27% to 63% for malignancy (21,23). In addition, organizing pneumonia is also a frequent complication of therapy in these patients (7% of our cohort) (Fig 3). Patients with HM are at particularly increased risk of invasive fungal infections (Fig 1). The diagnostic yield of biopsy for infections is generally lower than that for malignancies (range in the literature from 43% to 81% for infections vs up to 93% for malignancies) (24–26). Thus, patients in whom nonmalignant entities comprise a large fraction of lung diseases are likely to benefit more from the greater tissue yield of CNB.

Figure 1a:

Image in a 50-year-old man with history of diffuse large B-cell lymphoma and hematopoietic stem cell transplant who presented with fever and dry cough. (a) Axial CT chest section demonstrates left perihilar consolidation. CT-guided core-needle biopsy was performed with 20-gauge needle placed within consolidation. (b, c) Histopathology where methenamine silver stain and periodic acid-Schiff diastase stains highlight hyaline pauci septate broad hyphae compatible with zygomycetes.

Figure 2:

Image in a 77-year-old man with history of non-Hodgkin lymphoma. Enlarging right upper lobe nodule was seen on multiple CT scans. CT-guided fine-needle aspiration biopsy of nodule was performed with histopathology showing well-differentiated adenocarcinoma (not shown).

Figure 3:

Image in a 60-year-old man with history of acute myeloid leukemia. CT scan showed new mass in right lower lobe. CT-guided fine-needle aspiration biopsy was performed with 22-gauge needle. Histopathology demonstrated organizing pneumonia (not shown).

Figure 1b:

Image in a 50-year-old man with history of diffuse large B-cell lymphoma and hematopoietic stem cell transplant who presented with fever and dry cough. (a) Axial CT chest section demonstrates left perihilar consolidation. CT-guided core-needle biopsy was performed with 20-gauge needle placed within consolidation. (b, c) Histopathology where methenamine silver stain and periodic acid-Schiff diastase stains highlight hyaline pauci septate broad hyphae compatible with zygomycetes.

Figure 1c:

Image in a 50-year-old man with history of diffuse large B-cell lymphoma and hematopoietic stem cell transplant who presented with fever and dry cough. (a) Axial CT chest section demonstrates left perihilar consolidation. CT-guided core-needle biopsy was performed with 20-gauge needle placed within consolidation. (b, c) Histopathology where methenamine silver stain and periodic acid-Schiff diastase stains highlight hyaline pauci septate broad hyphae compatible with zygomycetes.

In our cohort, the occurrence rates of major complications were relatively low. The rates of chest tube placement in the CNB and FNAB groups (1.7% vs 1.9%) were in the lower end of the range reported in the literature (0.5%–31%) (27,28). Although patients with HMs are at increased risk of bleeding complications compared with the general population, particularly given frequent thrombocytopenia and dysfunctional platelets, our results for bleeding complications are comparable with studies in unselected patients. Rates of hemoptysis and hemothorax (2%–5%) in our cohort compare fairly with those published in the literature for these individual complications, which range between 4% and 6% (10,12,13,29). We also directly compared the complication rates of the study group with a control group of unselected lung biopsies from our own institution and found that complication rates were not statistically significantly higher in patients with HM than in the control group. It should be noted that many of the operators preferred FNAB over CNB for patients with HM, likely motivated by historic trends and the need to obtain microbiologic samples.

When complication rates were analyzed based on specific underlying diseases, there was a statistically significant difference only in hemoptysis rates (P = .048). No statistically significant difference was observed in pneumothorax rates (P = .93), chest tube rates (P = .52), or hemothorax rates (P = .16). However, given the multiple additional hypothesis tests performed on small subgroups, there is an increased risk of a type I error (false-positive significance). Therefore, these findings would have to be validated in an independent or a prospective cohort.

We did not find a statistically significant difference in the occurrence rates of major complications between FNAB-only and CNB (CNB + FNAB and CNB only) groups, similar to the results of previous studies in patients with non-HM (8,10,12). However, it should be noted that the 95% confidence intervals for complication rates were wide in our study owing to a small sample size. Therefore, it is possible that there is a true difference that our study is underpowered to detect. We did see a trend toward higher rates of hemoptysis and hemothorax in patients with lower platelet counts irrespective of the biopsy technique, but this did not reach statistical significance. Of note, there was a statistically significant increased risk of bleeding complications in patients after HSCT (12% vs 2%); it is not clear why this is the case, although it is possible that these patients may have dysfunctional platelets. There was one death in our cohort of patients (in the CNB + FNAB group in a patient after HSCT), giving a mortality rate of 0.6% (one of 166), which is comparable to mortality rates in the literature, ranging from 0.15% to 0.5% (30–32).

In our cohort, 51% of patients who underwent diagnostic biopsies experienced a change in clinical management versus 21% of patients who underwent nondiagnostic biopsies, demonstrating that obtaining a specific diagnosis is associated with a clinically relevant management change. These results support similar findings demonstrated in previous studies on lung biopsies in immunocompromised patients (18,21,23).

This study had several limitations. It was retrospective, and the choice of CNB versus FNAB and size of the needle was determined by individual operators based on operator preference, patient risk factors, and lesion characteristics, rather than a standardized protocol. The group comparisons are, therefore, also partially confounded with the effects of competing risk factors. Of note, to mitigate this fact, we did perform a propensity risk score–matched analysis that yielded similar results to our unadjusted cohort. In addition, the number of needle passes, location, and size of the lesion, which may influence both complication rates and yield, were not included in the analysis. In addition, quantification of degree of hemoptysis and pneumothorax could not be performed owing to sometimes limited data in the medical record; however, we did record chest tube placement rates. Our study did not assess the survival benefit from any treatment changes instituted after biopsy. Additionally, we did not assess any further clinical or interventional work-up after nondiagnostic biopsies.

In conclusion, we found that CNB is an effective technique for CT-guided lung biopsies in patients with HMs, with a greater diagnostic yield compared with FNAB and higher although not statistically significant increased risk of biopsy-related bleeding complications or pneumothorax. Because our study was not adequately powered to detect infrequent but catastrophic outcomes such as death, larger studies are needed to investigate such complications.