Utility of ¹⁸F FDG PET/CT in Hemophagocytic Lymphohistiocytosis

Introduction

Hemophagocytic lymphohistiocytosis (HLH) is a potentially fatal systemic inflammatory syndrome caused by immune dysregulation. Its recent global incidence and prevalence are unknown, but it has regional variations with tropical climate countries having a higher risk of HLH.[1234567] It is either familial (around 25% cases) or acquired secondary to varied etiologies ranging from infections, autoimmune disorders, and malignancies.[1] The diagnosis of HLH is complex and involves the fulfillment of five out of eight HLH-2004 criteria comprising various pathological and biochemical parameters for a definite diagnosis.[8] However, these parameters were derived from a pediatric cohort and lacked some of the commonly observed clinical features of HLH. Thus, the H-score was developed using an adult cohort, and a cutoff score >169 had high diagnostic accuracy for HLH.[91011] Both criteria have shown high diagnostic accuracy for detecting HLH in adults with some studies showing superiority of the H-score.[1011] Hemophagocytosis, although present during the disease course at some stage, is not mandatory for diagnosing HLH and may be seen in sepsis or after transfusion.[912] The diagnosis and etiology of HLH often remain uncertain during the lifetime of the patient due to delayed clinical recognition, high mortality in the absence of treatment, and a lack of widespread availability of genetic and laboratory testing facilities, especially in low- and middle-income countries.[13] HLH has a multiorgan and multisystemic involvement with varied presentations, often characterized by nonspecific imaging findings associated with the inciting pathology.[1415] Moreover, it has an overlapping clinical picture with sepsis, making diagnosis difficult and challenging.[16] ¹⁸F fluorodeoxyglucose (FDG) positron emission tomography (PET) is a highly sensitive imaging modality with high diagnostic performance in infections, inflammatory disorders, and malignancies.[171819] However, the literature on its utility in HLH is limited. This study aims to evaluate the role of ¹⁸F FDG PET/computed tomography (CT) in HLH.

Materials and Methods

A retrospective evaluation of a prospectively maintained electronic registry was done to identify all patients with HLH who were referred for a whole-body ¹⁸F FDG PET/CT from 2010 to April 2024. The diagnosis of HLH was based on the H-score involving various clinical (organomegaly, fever, and immunosuppression), histopathological (bone marrow [BM] aspirate), and laboratory (complete blood count, liver function tests, lipid profile, coagulation profile, and serum ferritin) investigations.[9] The PET/CT findings were independently reviewed by two experienced nuclear medicine physicians with more than 10 years of experience, to identify the potential cause for HLH. Treatment naïve children who were primary HLH suspects and evaluated with FDG PET/CT to rule out other etiologies were also included in the study. All patients with HLH who underwent treatment before PET/CT were excluded from the study. A positive PET finding was considered focal FDG uptake at sites other than physiological sites. For organs such as the spleen and BM, diffuse FDG uptake greater than physiological liver uptake was considered positive. The final diagnosis for the etiology of HLH was confirmed using various laboratory (serology and culture), genetic (next-generation sequencing), and histopathological (fine-needle aspiration and biopsy) tests. The diagnosis of autoimmune disease was based on an amalgamation of clinical and serological findings based on standard diagnostic criteria. The maximum standardized uptake value (SUV_max) of the liver, spleen, and BM (measured at L4 vertebral body) were measured, and the spleen-to-liver ratio (SLR) and bone-to-liver ratio (BLR) were calculated. These PET parameters were correlated with various hematological parameters, biochemical parameters, and BM findings. SPSS v. 26.0 (IBM, Armonk, NY, USA) was used for statistical analysis. P < 0.05 was considered statistically significant.

Results

Causes of hemophagocytic lymphohistiocytosis

The most common secondary cause of HLH was infection (8/24, 33%), followed by malignancy (7/24, 29%) and autoimmune disease (3/24, 13%). The cause was unknown in one patient. Out of eight patients with infection [Figure 1], six had viral infections (three with viral hepatitis and three with Epstein–Barr virus [EBV]) and two had tuberculosis (TB). Both patients with TB had hepatomegaly, splenic uptake > liver uptake, lymph node and lung involvement, and basal ganglia hypermetabolism, whereas one had BM involvement and splenic lesions without splenomegaly. All patients with EBV infection had hepatomegaly, BM involvement, and basal ganglia hypermetabolism, whereas two patients had splenomegaly and lymph node involvement and one had lung involvement. Only two patients with viral hepatitis had hepatomegaly and lymph node involvement, one had splenomegaly and lung involvement, and none had BM involvement or high splenic uptake. The median SUV_max of the hottest lymph nodes was significantly higher in those with malignancy versus viral infection (14.0 vs. 2.5, P = 0.015) while no significant difference was observed between malignancy and TB (14 vs. 9.2, P = 0.7).

Close

Figure 1 Maximum intensity projection images of ¹⁸F fluorodeoxyglucose positron emission tomography/computed tomography of infection-associated hemophagocytic lymphohistiocytosis patients (a) A 47-year-old male with disseminated tuberculosis (TB), (b) A 82-year-old male with disseminated TB, (c) A 22-year-old male with hepatitis A virus infection, (d) A 18-year-old female with Epstein–Barr virus (EBV) infection presenting as ileocecal thickening and ileocolic lymph nodes, (e) A 5-year-old female with EBV infection presenting with lymph nodes on both sides of the diaphragm. EBV: Epstein–Barr virus, TB: Tuberculosis

Export to PPT

Out of seven patients with malignancy [Figure 2], four had non-Hodgkin’s lymphoma while three had HL. All of them had hepatomegaly with two (29%) having liver lesions, five (71%) having splenomegaly with splenic uptake > liver uptake, and four (57%) having splenic lesions. Five patients (71%) had lymph node involvement and six (86%) had BM involvement with four having focal on diffuse involvement and two having diffuse BM uptake. Three patients had basal ganglia hypermetabolism, one patient had lung, pancreas, renal, muscle, and subcutaneous lesions, and another patient had vanishing bile duct syndrome with pleural effusion and ground-glass opacities.

Close

Figure 2 Maximum intensity projection images of ¹⁸F fluorodeoxyglucose positron emission tomography/computed tomography of malignancy-associated hemophagocytic lymphohistiocytosis patients (a) A 42-year-old male with B-non-Hodgkin lymphoma, (b) A 22-year-old male with HL, (c) A 9-year-old child with HL, (d) A 21-year-old male with diffuse large B-cell lymphoma, and (e) A 19-year-old male with B-cell acute lymphoblastic leukemia. B-NHL: B-non-Hodgkin lymphoma, DLBCL: Diffuse large B-cell lymphoma, HL: Hodgkin's lymphoma

Export to PPT

Three patients with autoimmune involvement had adult-onset Still’s disease (AOSD) with all having splenomegaly (with two having splenic uptake > liver uptake), two each having hepatomegaly, lymphadenopathy, BM involvement, and serositis, and one having basal ganglia hypermetabolism. The imaging findings are summarized in Table 2. Overall, eight patients had CNS findings on imaging in the form of basal ganglia hypermetabolism. The median C-reactive protein (CRP) levels of patients with this finding were higher than those without Basal ganglia (BG) hypermetabolism without reaching a significant difference (40.9 vs. 14.9, P = 0.5).

Table 2 Fluorodeoxyglucose-18 positron emission tomography/computed tomography findings according to the cause of hemophagocytic lymphohistiocytosis

Cause of HLH	Hepatomegaly	Splenomegaly	Spleen uptake > liver uptake	BM involvement	Lymph node involvement	Basal ganglia hypermetabolism	Lung involvement
Malignancy (n=7)	7 (100)	5 (71), splenic lesions in 4	5 (71)	6 (86%), focal on diffuse (4), diffuse (2)	5 (71)	3 (43)	2 (29)
TB (n=2)	2 (100)	1 (50), splenic lesions in 1 (50)	2 (100)	1 (50)	2 (100)	2 (100)	2 (100)
EBV infection (n=3)	3 (100)	2 (67)	1 (33)	3 (100)	2 (67)	3 (100)	1 (33)
Viral hepatitis (n=3)	2 (67)	1 (33)	0	0	2 (67)	1 (33)	1 (33)
Autoimmune (n=3)	2 (67)	3 (100)	2 (67)	2 (67)	2 (67)	1 (33)	1 (33)
Primary HLH (n=5)	5 (100)	4 (80), 1 with splenic lesions	3 (60)	3 (60)	4 (80)	0	1 (20)

HLH: Hemophagocytic lymphohistiocytosis, BM: Bone marrow, EBV: Epstein–Barr virus, TB: Tuberculosis

After excluding secondary causes, subsequent testing revealed primary HLH in five patients (21%) [Figure 3]. FDG PET/CT was positive in all five patients with primary HLH. All had hepatomegaly, four had splenomegaly (three having high splenic uptake and one with splenic lesions) and lymph node involvement, and three patients had BM involvement. Two patients had X-linked lymphoproliferative syndrome with one patient having SH2D1A gene deletion (X-linked lymphoproliferative syndrome type I) and acute EBV infection. A 29-year-old male with HLH presented with multiple hypopigmented and hyperpigmented patches on his face and limbs along with numbness and paresthesia in his upper and lower limbs. He also had hypopigmented patches of hair. FDG PET revealed multiple FDG avid cutaneous and subcutaneous nodularities, intramuscular deposits, hepatosplenomegaly, and marrow hypermetabolism. ¹⁸F FDG PET-/CT-guided biopsy from muscular lesion revealed lepromatous leprosy. He had a family history of premature death in two family members due to prolonged fever. Next-generation sequencing of DNA revealed a homozygous deleterious mutation in the RAB27A gene (Griscelli syndrome type 2) predisposing to HLH development.

Close

Figure 3 Maximum intensity projection images of ¹⁸F fluorodeoxyglucose positron emission tomography of patients with primary hemophagocytic lymphohistiocytosis (HLH) (a) A 4-year-old male with X-linked lymphoproliferative disease (XLPD) and Epstein–Barr virus (EBV) infection, (b) A 1-year-old male child with XLPD, (c) A 3-year-old male child with primary HLH and EBV infection, (d) A 29-year-old male with Griscelli syndrome type II presenting with disseminated lepromatous leprosy, and (e) 13-year-old male child with primary HLH

Export to PPT

In eleven patients (46%), ¹⁸F FDG PET findings were used for guided targeted histopathological sampling from the most hypermetabolic sites on imaging. In all cases, sampling helped to reach the final diagnosis or to eliminate malignant causes.

Follow-up PET/CT was available in five of seven patients with lymphoma as per standard guidelines; one patient each with TB, AOSD, and suspected malignancy (where follow-up PET and PET-guided biopsy led to the final diagnosis of primary HLH).

Semiquantitative positron emission tomography analysis

There was no significant difference in SUV_max (4.0 vs. 3.5, P = 0.6) and BLR (1.475 vs. 1.514, P = 0.4) in patients with and without evidence of HLH on BM sampling, but a significant difference in median marrow SUV_max (5.1 vs. 3.2, P = 0.019) and BLR (1.58 vs. 0.82, P = 0.043) was observed in patients with hypercellular versus normocellular/hypocellular marrow. A significant positive correlation was observed between splenic uptake (SUV_max) and BM uptake (SUV_max) (r = 0.501, P = 0.013) and between BLR and SLR (r = 0.623, P = 0.001) [Figure 4]. CRP levels significantly correlated with BLR (r = 0.731, P = 0.001) and SLR (r = 0.594, P = 0.015). A significant difference was noted among the BLR (P = 0.039) of various etiologies. Post hoc analysis revealed a significant difference between BLR of malignancy and viral infection (1.65 vs. 1.0, P = 0.007). No correlation was observed between other hematological and PET parameters.

Close

Figure 4 Scatter plots showing the correlation between (a) splenic uptake and bone marrow uptake, (b) spleen-to-liver ratio (SLR) and bone-to-liver ratio (BLR), (c) C-reactive protein (CRP) and BLR, and (d) CRP and SLR. SLR: Spleen-to-liver ratio, BLR: Bone-to-liver ratio

Export to PPT

Discussion

HLH is a fatal systemic hyperinflammatory syndrome which requires early detection and management. Infections, particularly viral infections are the most common cause of secondary HLH, followed by other infections such as TB, malignancy, and autoimmune disorders which was also true for our cohort.[132021] The diagnosis of HLH requires a multitude of clinical–biochemical tests, and despite a comprehensive workup, a definitive diagnosis and cause often remain unclear due to close mimics. The role of imaging is unclear in diagnosing HLH and evaluating its causes. It presents with a range of nonspecific imaging findings commonly involving the CNS, pulmonary, and abdominal organs.[14] Rather than diagnosing HLH, imaging can particularly help identify its potential cause and guide targeted sampling for earlier diagnosis and treatment. Hepatomegaly was seen in 92% of patients in this cohort, whereas splenomegaly was observed in 67% of the patients. Another common finding was the presence of high splenic and BM uptake greater than the physiological liver uptake. The positive correlation of acute-phase reactants like CRP with BLR and SLR may indicate the systemic hyperinflammatory state associated with HLH. High FDG uptake in the BM and spleen is associated with high serum cytokine levels, and the positive correlation between spleen and BM uptake and BLR and SLR also indicate the concordant hematopoietic cell activity in both organ systems.[222324] The pattern of FDG uptake in the BM and spleen can help differentiate a reactive uptake due to systemic inflammation from disease involvement. A focal on diffuse uptake pattern in the BM is confirmatory for disease involvement, whereas a diffuse uptake pattern is nonspecific and requires histopathological confirmation for distinction. Diffuse uptake in the BM and spleen represents indirect signs of inflammation and may be caused by multiple etiologies such as infection, autoimmune disorders, or malignancy.[25] Distinction among these etiologies solely based on the intensity of diffuse uptake may be inadequate, and histopathological confirmation is crucial. This finding, although nonspecific, has important implications in HLH. The intensity of FDG uptake correlates with disease activity, and severity and has a prognostic role.[24] It can also be used to assess response to treatment. Our analysis also revealed a significantly higher BLR in HLH patients with malignancy than those with viral infection (1.65 vs. 1.0, P = 0.007). Similar results were observed in previous studies where metabolic uptake of organs in malignancy-associated HLH was significantly higher than in patients with infection-associated or rheumatosis-associated HLH.[26] In addition, EBV infection may also present with lymph node involvement on both sides of the diaphragm mimicking lymphoma and may cause false-positive interpretation on ¹⁸F FDG PET.[2728] However, the FDG avidity of lymph nodes in EBV infection is lower than those in lymphoma, and the appearance of extranodal hypermetabolic lesions is associated with lymphoma involvement.[29] In our study too, the SUV_max of involved lymph nodes was significantly higher in those HLH patients with malignancy than viral infection (14.0 vs. 2.5, P = 0.015), whereas no significant difference was observed between malignancy and TB (14 vs. 9.2, P = 0.7). TB often mimics malignancy in terms of metabolic uptake, and it is difficult to distinguish between them solely based on metabolic signature on ¹⁸F FDG PET.[30] Systemic cytokines have a wide array of manifestations on ¹⁸F FDG PET including basal ganglia hypermetabolism which may present without neurological symptoms.[31] In our cohort, those patients with basal ganglia hypermetabolism had higher median CRP levels than those without this finding without reaching significant values. It may indicate a positive correlation between serum cytokine levels and systemic inflammatory activity; however, it requires evaluation in a larger cohort for validation. Previous studies have shown a positive correlation of total leukocyte count and neutrophil counts with SLR; however, no significant correlation was observed between imaging and hematological parameters in our study.[2224] The presence of HLH on BM sampling does not necessitate a high BM FDG uptake or a higher BLR; however, the median BM uptake and BLR were significantly higher in hypercellular BM compared to a normocellular or hypocellular marrow confirming previously published findings.[32] However, no previous studies have correlated the presence of HLH on BM biopsy with ¹⁸F FDG PET findings.

Although patients with HLH have different causes and varying imaging findings, the majority of them present with pyrexia of unknown origin. The recent consensus document on the role of ¹⁸F FGD PET in fever and inflammation of unknown origin recommends FDG PET/CT over conventional imaging due to its high diagnostic yield, whole-body acquisition, and guiding targeted biopsy from most suspicious and active sites.[25] Rather than defining diagnostic accuracy parameters such as sensitivity and specificity, in diseases like HLH having a multitude of causes, other parameters such as diagnostic yield and helpfulness of imaging modality gain more importance.[25] The nonspecific nature of ¹⁸F FDG PET/CT may be considered an advantage due to the broad spectrum of differential diagnoses in Fever of unknown origin/inflammation of unknown origin (FUO/IUO) or HLH.[25] In our cohort, ¹⁸F FDG PET/CT had a high diagnostic yield of 92%, indicating its potential use in early diagnosis and guiding management. Despite this high diagnostic yield, false-positive uptake due to nonspecific findings and false-negative scans due to treatment-related sequelae must be incorporated into the final diagnostic interpretation.

In a resource-limited setting with lower availability and accessibility to genetic testing, the diagnosis of primary HLH is made after excluding all secondary causes of HLH and correlation with family history and age of the patient.[33] A highly sensitive imaging modality like FDG PET/CT is useful to exclude the causes of secondary HLH due to its high negative predictive value. In addition, due to immunodeficiency, primary HLH patients are prone to hematolymphoid malignancies and infections, and ¹⁸F FDG PET/CT plays a crucial role in early diagnosis and management. Some genetic mutations (SH2D1A) make the individual prone to EBV infections, increasing the risk of developing EBV-associated lymphomas and chronic infections.[3435] EBV has a complex relationship with HLH. It can be a trigger for primary HLH or can cause secondary HLH either in a nonneoplastic or neoplastic background.[36] Mutations in other genes like RAB27A cause Griscelli syndrome type 2, characterized by partial albinism and immunodeficiency due to reduced and dysfunctional T and NK cells, predisposing to recurrent infections and HLH, like in one patient of our cohort.[37] An amalgamation of imaging findings on ¹⁸F FDG PET/CT and guided biopsy along with clinical findings and family history were crucial in clinching the diagnosis and cause of HLH. There is a dearth of literature on the utility of ¹⁸F FDG PET/CT in primary HLH.

In our cohort, using ¹⁸F FDG PET findings to guide targeted histopathological sampling helped to reach the final diagnosis in most patients. It was also helpful in narrowing the cause of secondary HLH, especially in diagnosing or eliminating malignancy.

Our study is limited by its retrospective nature and small sample size. Despite these limitations, this is one of the largest cohorts of HLH patients describing FDG PET/CT findings. In addition, due to absent recommendations and guidelines for ¹⁸F FDG PET/CT in HLH, this cohort is a collection of heterogeneous groups of patients. Prospective large stratified sample studies based on multiple causes of HLH are required to gather more evidence on the role of ¹⁸F FDG PET/CT findings in HLH.

Conclusion

¹⁸F FDG PET/CT is a sensitive modality for evaluating the causes of HLH and helps in guiding targeted sampling for final diagnosis. However, it is a supportive modality that can be used in conjunction with the clinical and laboratory parameters and may be utilized earlier in the diagnostic algorithm of HLH.