Granulomatous Lymphocytic Interstitial Lung Disease (GL-ILD): CT Imaging Features and Clinical Findings to Expedite a Timely Diagnosis

Background: Establishing a diagnosis of Granulomatous Lymphocytic-Interstitial Lung Disease (GL-ILD) is dicult due to the overlapping CT imaging and histologic features, and are often confused with infection, sarcoidosis and/or follicular bronchiolitis. Research Question: Identify specic CT imaging features and clinical characteristics to suggest a condent diagnosis of Granulomatous Lymphocytic-Interstitial Lung Disease (GL-ILD). Study Design and Methods: IRB-approved retrospective case series study involving the review of the electronic medical record and CT chest imaging of eight patients with GL-ILD. Results: Bronchocentric airway-centered lower lobe predominant (part solid nodules more common than ground glass or solid nodules) ranging from 5 to 10 mm in size were found consistently in a majority of the eight patients with GL-ILD. Interpretation: In combination with the above pulmonary nodular pattern, mediastinal and hilar lymphadenopathy in conjunction, either splenomegaly or splenectomy and/or abdominal lymphadenopathy are excellent CT imaging features to prospectively suggest the diagnosis of GL-ILD in a patient with CVID.

Interpretation: In combination with the above pulmonary nodular pattern, mediastinal and hilar lymphadenopathy in conjunction, either splenomegaly or splenectomy and/or abdominal lymphadenopathy are excellent CT imaging features to prospectively suggest the diagnosis of GL-ILD in a patient with CVID.

Background
Granulomatous lymphocytic interstitial lung disease (GL-ILD) is an interstitial lung disease (ILD) that occurs in patients with common variable immunode ciency (CVID) and is associated with a lymphocytic in ltrate and/or granulomatous in ammation in the lung, and diagnosed when other possible conditions have been excluded [1]. The non-infectious pulmonary complications of CVID include organizing pneumonia, follicular bronchiolitis, GL-ILD, malignancy (lymphoma), obstructive lung disease secondary to bronchiolitis obliterans (BO), bronchiectasis/bronchiolectasis and pulmonary nodules [2,3]. It is challenging to distinguish the infectious pulmonary manifestations and their associated sequela from noninfectious in ammatory processes as a result of overlapping clinical, radiological and pathological features.
A con dent diagnosis of GL-ILD often requires a histological evaluation [4] combined with cultures and special stains to exclude infection [4,5]. However, surgical lung biopsies in patients with CVID can be associated with an increased risk of decompensation, infection and surgical complications [6]. It is therefore crucial to recognize and suggest GL-ILD in CVID patients with the appropriate CT imaging pattern to allow for early diagnosis and treatment. GL-ILD can be confused with other common and uncommon diseases like sarcoidosis and/or (follicular) bronchiolitis both histologically and on CT chest imaging [5][6][7][8]. Characterization of CT imaging features that allow the radiologist to suggest a con dent early diagnosis can reduce this diagnostic dilemma. Recognizing and de ning the clinicopathological features, as well as the radiological features, of this condition will help improve early identi cation, con dent diagnosis and treatment of this multi-systemic disease [9].
Presence of solid nodules < 3 cm, semisolid nodules, pure ground-glass opacities, enlarged thoracic (hilar and/or mediastinal) lymph nodes, and splenomegaly have been suggested as typical in GL-ILD but none were consistently rated as necessary to make the diagnosis [1,10]. There is a need for better de ning the radiological features to avoid the risk of complications from surgical biopsies in this high-risk population.
To date, there is no clear consensus on the CT imaging features or their distribution for GL-ILD. The purpose of this retrospective case series was to evaluate our cohort of eight patients with CVID that developed a GL-ILD pattern on CT in conjunction with clinical and/or pathologic diagnosis of GL-ILD to establish speci c CT imaging features to allow the radiologist to suggest a presumptive diagnosis of GL-ILD, possibly obviating a potential surgical lung biopsy.

Study population -
Approval was obtained to review the electronic medical record for retrospective case series by the institutional review board of Emory University and was performed in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki). The electronic medical record was used to search for patients with a diagnosis of CVID made by an immunologist from 1995 to 2018 and a con rmed diagnosis of GL-ILD by a multidisciplinary discussion (MDD) approach involving the interstitial lung disease (ILD) team comprised of ILD pulmonologists, immunologist, thoracic radiologist and pathologists ( Table 1). The electronic medical record was reviewed of those with CVID and a cohort of patients with a con rmed diagnosis of CVID and chest CT imaging from 1995 to 2018 (Table 1). All pertinent clinical information was retrieved from the electronic medical record, including demographics, pulse oximetry readings, symptoms, pulmonary function test (PFTs) data, histopathology and laboratory results.

CT evaluation -
The CT scans used for scoring was the most recent CT chest; however, previous CT scans were referred to when needed for comparison. The presence of an active infection was excluded by chart review, cultures and special stains. CT chest imaging was performed on a variety of scanners from several different manufacturers within our institution: GE Lightspeed VCT spiral acquisition, 120 to 140 KVP, slice

Method -
Two thoracic radiologists (EB with 12 years and BL with 10 years of experience in thoracic radiology) independently reviewed CT scans and scored according to a scoring sheet agreed upon by the study team prior to reviewing the CT scans. Disagreements were resolved by consensus. Data was collected on the following parameters: the presence, absence, attenuation/density, size and distribution of pulmonary nodules; the presence, absence and distribution of bronchiectasis; the presence, absence, axial distribution, size of ground glass opacities; presence or absence of bronchial wall thickening, air trapping/mosaic attenuation, pleural effusions, consolidations, hepatomegaly, splenomegaly, abdominal lymphadenopathy and hilar/mediastinal lymphadenopathy.

Results
Our case series had eight patients, all non-Hispanic white 5 females and 3 males, ranging in age from 30 to 66; ve of the eight patients had immune thrombocytopenic purpura (ITP) ( Table 1). Five out of the eight patients had symptoms at presentation. Only two of the patients were ex-smokers (case 3 -5 pack years, case 8 -20 pack years). Case 3 and 7 had hypoxia requiring oxygen therapy. Case 3 was on treatment for a B cell lymphoma diagnosed by subcarinal lymph node biopsy and ow cytometry ( Figure  1E and 1G). Three patients were on rituximab for treatment of GL-ILD. Seven of the eight patients had pulmonary function testing (PFT) and ve patients have had full PFT with lung volume testing. The most predominant PFT pattern was a reduced diffusion capacity of carbon monoxide (DLCO) seen in four patients. Restrictive lung functions were seen in three patients. Air ow limitation/obstructive pattern was not seen in any of the patients. (Table 1)   Table 2 summarizes imaging data with regard to nodule distribution and density, and the presence or absence of bronchiectasis, reticulation, splenomegaly, and lymphadenopathy. The distribution of nodules is bronchocentric with a lower lobe predominance ( Table 2). All the lobes had more than 10 nodules bilaterally in all 8 patients. The largest nodule size was > 10 mm in all the lobes. Overall, the most common nodule density was part solid (Figures 1 and 2) but solid and ground glass nodules were present in all except for 1 patient who had predominantly part solid nodules. Bronchiectasis was not a common nding and when present, was very mild, cylindrical and lower lobe predominant. Mild to moderate smooth bronchial/bronchiolar wall thickening was present in all eight patients ( Figure 1B). Lower lobe predominant patchy ground glass opacities were seen on the CT chest in seven patients ( Figure 2). Three patients had mild lobular geographic areas of mosaic attenuation representing air trapping, such as, patient 4 which demonstrated lower lobe predominant small part solid, solid and ground glass nodules ( Figure 3A), representing follicular bronchiolitis and scattered poorly-formed non-necrotizing granulomas ( Figure 3B). Lower lobe predominant ne reticulations were present in four of the eight patients. None of the patients had consolidation or pleural effusion. Five patients had splenomegaly ( Figure 1F) while the other three patients had splenectomies due to previous splenomegaly; none of the patients had hepatomegaly. Seven patients had abdominal lymphadenopathy. Mediastinal lymphadenopathy ( Figure  1 D and E) was seen in all eight patients and hilar lymphadenopathy ( Figure 1E) in seven patients. Figure  4 demonstrates the apical to basal gradient of the solid, part-solid and ground glass nodules, most prominent in the lower lobes with intermixed patchy areas of ground glass opacities. An early con dent diagnosis allowed treatment to start early with a positive response to therapy as demonstrated in Figure   5.

Discussion
CVID is a primary immunode ciency characterized by decreased IgG two standard deviations below the lower limits of normal and reduced IgA and/or IgM with impaired speci c antibody response measured via vaccine titers and variable T cell defects [3], that cause recurrent infections, particularly upper and lower respiratory tract infections [2]. GL-ILD is a non-infectious complication of CVID that manifests various lung ndings on CT imaging that have not been previously well characterized with respect to morphology and distribution. Additionally, there are several histological patterns including granulomatous in ammation, lymphoid hyperplasia/lymphocytic interstitial pneumonia (LIP) and follicular bronchiolitis [2,9,11] that can be confusing to the pathologist. A retrospective series that examined lung pathology in 34 patients with CVID demonstrated benign lymphoid proliferation and granulomas in 23 patients who did not have a clinical and radiological diagnosis of GL-ILD [5]. Over-diagnosing GL-ILD in patients histologically without tight clinical, historical and radiological criteria and multidisciplinary agreement does not provide useful information in managing patients. Patients with CVID may have other pulmonary manifestations such a bronchiectasis, necrotizing granulomas from infection, lymphocytic intestinal pneumonia (LIP) from coexisting autoimmune disease [5,12], scarring and pneumonia that may overlap with GL-ILD [13]. Some of the CT imaging features associated with GL-ILD are common to sarcoidosis [14], and are thus often misdiagnosed as sarcoidosis, especially histologically [6,7,14]. Histology may not be the "gold standard" in diagnosing GL-ILD rather a multidisciplinary ILD conference is the most optimal method to obtain a diagnosis of GL-ILD with the radiologist raising GL-ILD in the differential diagnosis. In centers with multidisciplinary ILD teams, a lung biopsy can be reserved to resolve disagreements and ambiguity.
Although GL-ILD suggests a pulmonary process, it actually encompasses a multi-systemic granulomatous in ammatory disease that may affect the liver, spleen, lymph nodes and any organ/organ system [15]. Due to the lymphocytic dysregulation in GL-ILD, patients are at high risk of lymphoma [16].
Several theories have been proposed to explain the pathogenesis of GL-ILD in CVID, including dysfunctional antigen handling (due to impaired T cell function) and aberrant immune response to viruses (potentially HHV-8). Due to improved life expectancy and reduced rate of infection with immunoglobulin replacement therapy, non-infectious manifestations of CVID such as GL-ILD are becoming more emergent [6,11]. GL-ILD treatment requires immunomodulatory therapy (rituximab, azathioprine), as well as high dose immunoglobulin replacement therapy [13,17] to reduce the disease progression and potentially reverse some of the radiological disease [9]. Some patients diagnosed with GL-ILD with normal lung functions and mild disease may not need treatment unless there is radiological and clinical progression [18].
The main goal of our case series was to identify predominant clinical and imaging ndings in this cohort of GL-ILD patients that can be used to identify patients with CVID who have developed GL-ILD. The lower lobe predominant ground glass opacities (GGOs) (7 out of the 8 patients) were a nding more common in prior CT scans that tend to evolve into solid nodules in the subsequent scans. While longitudinal imaging studies are not available, it can be theorized that these GGOs are nodules in evolution, based on their distribution. There was 98% agreement between the two independent chest radiologists for the individual radiological ndings of bronchocentric nodules with an apical to basal gradient, most numerous in the lower lobes.
When a PFT abnormality was present, the predominant nding was a moderate to severe reduction in the diffusion capacity, two patients who had severe impairment of diffusion capacity had hypoxia (Table 1).
Restrictive lung functions are the predominant nding in GL-ILD in keeping with previous studies and case reports [11,12,18]; no patient had air ow limitation/ obstructive physiology that would implicate airway involvement [12]. Splenomegaly and ITP were important predictors of GL-ILD in CVID patients that were identi ed by the case control study by Hartano et al (69%) [6]. Splenomegaly was also an important clinical nding towards establishing a diagnosis of GL-ILD in the BLS/UKPIN statement [1], ve patients in our series demonstrated splenomegaly while the other three patients had a splenectomy ( Table 2).
Features of ITP were present in ve patients. Mild to moderate cylindrical bronchiectasis that was found in six patients is representative of the sequel of recurrent infections in CVID rather than a feature of GL-

ILD.
Limitations of the GL-ILD case series is the small sample size of eight, but representative of the prevalence of this rare disease and the strict case selection by the multidisciplinary ILD team. Additional cohorts in a multicenter approach would be necessary to validate these ndings. This study has the restrictions of a retrospective chart review and our data is limited by the availability of test results and documentation in the chart. One of the strengths that stand out in this study is the in-depth explanation of the radiological ndings with CT imaging examples from our cohort.
Future studies should focus on multi-center database approach to increase the sample size of this rare disease. HRCT and better de ned radiological criteria are replacing the need for biopsy in the eld of ILD. GL-ILD has distinct radiological ndings that can be useful in centers with multidisciplinary ILD teams.

Interpretation
Bronchocentric airway-centered lower lobe predominant (part solid nodules more common than ground glass or solid nodules) ranging from 5 to 10 mm in size were found consistently in a majority of the eight patients con rmed with GL-ILD. In combination with the above pulmonary nodular pattern, mediastinal and hilar lymphadenopathy in conjunction, either splenomegaly or splenectomy and/or abdominal lymphadenopathy are excellent CT imaging features to prospectively suggest the diagnosis of GL-ILD in a patient with CVID.

Tables
Due to technical limitations, tables are only available as a download in the Supplemental Files section.