Disseminated tuberculosis masquerading as a neuroendocrine tumour

Abstract

We describe the case of a 61-year-old man from the Dominican Republic admitted with diarrhoea, fevers and weight loss who was found to have lab studies and imaging (including radiolabeled somatostatin positron emission tomography/CT scan) initially consistent with a metastatic neuroendocrine tumour. However, after weeks of workup and multiple inconclusive biopsies, he was diagnosed with disseminated extrapulmonary tuberculosis. Here we examine the data for neuroendocrine tumour and tuberculosis labs and imaging to delineate where these studies overlap. We also analyse the biases and pitfalls in this case that led to a protracted diagnosis.

Background

Tuberculosis (TB) is the leading cause of death from infectious disease worldwide with approximately 1.7 million deaths reported in 2016.¹ The greatest burden of new infections occurs in the developing world; Europe and North America account for less than 6% of incident cases. In the USA, rates of TB continue to decline with an incidence rate of 2.9 per 100 000 in 2016.² Consistent with these data, in the USA, rates of TB among foreign-born persons is 15 times the rate of US-born persons.

More than 80% of cases of active TB in the USA can be attributed to reactivation of latent TB infection (LTBI) rather than to recent transmission.³ Many risk factors associated with immunosuppression promote the reactivation of LTBI, most notably HIV/AIDS, organ transplantation, end-stage renal disease and diabetes mellitus.⁴ Extrapulmonary TB (EPTB) occurs in 15%–20% of cases and commonly involve the pleura, lymph nodes, osteoarticular sites, gastrointestinal (GI) and genitourinary (GU) tracts.⁵

EPTB is known to mimic many conditions including malignancies. Neuroendocrine tumours (NETs) are uncommon malignancies that arise from the hormonal or peripheral nervous system and are rarely implicated in cases of EPTB. Certain NETs can secrete serotonin and lead to carcinoid syndrome, which is characterised by a constellation of symptoms including diarrhoea, fevers, flushing and right-sided heart failure.⁶ A high index of suspicion is required to diagnose these tumours. The main diagnostic modalities include laboratory markers such as chromogranin A and neuron-specific enolase as well as imaging with radiolabelled somatostatin (⁶⁸Ga DOTATATE) positron emission tomography (PET)/CT scan.

We report the case of an individual with a history of LTBI and multiple risk factors for reactivation who was found to have an atypical case of disseminated EPTB mimicking a NET in presentation, lab results and imaging.

Case presentation

A 61-year-old man from the Dominican Republic with a history of latent TB treated with isoniazid, type 2 diabetes and end-stage renal disease (ESRD) on haemodialysis was admitted for recurrent fevers, watery diarrhoea with 3–6 episodes daily and abdominal distension for the past 3 months. In addition, he reported a weight loss of 30–35 pounds over this time period.

At the time of presentation, he appeared cachectic, with a prominently distended abdomen and a positive fluid wave. Aspartate aminotransferase and alanine aminotransferase were within normal limits, while alkaline phosphatase was elevated at 766 IU/L (normal 41–133 IU/L) and gamma-glutamyl transferase was elevated at 398 U/L (normal <73 U/L). C-reactive protein and erythrocyte sedimentation rate were both elevated.

He underwent a diagnostic paracentesis on admission which demonstrated a low serum-ascites albumin gradient (SAAG), 200 nucleated cells with 44% neutrophils, negative acid fast bacilli (AFB) smear and negative bacterial, fungal and AFB cultures. Other infectious workup including bacterial blood cultures, HIV serology, hepatitis panel and a Clostridium difficile toxin assay yielded negative results.

CT imaging demonstrated scattered subcentimetre pulmonary nodules, diffuse lymphadenopathy, multiple small hypoenhancing liver lesions, large volume ascites and areas of peritoneal thickening described as carcinomatosis (figure 1), suspicious for a metastatic malignancy. An oncology consultation was obtained and initial cancer markers for abdominal/pelvic malignancy including cancer antigen 19-9, prostate-specific antigen and carcinoembryonic antigen were within normal limits.

Figure 1.

(A) Circled are multiple small hypoenhancing liver lesions. Large volume ascites also apparent. (B) Arrow pointing to area of peritoneal thickening described originally as carcinomatosis.

Outcome and follow-up

Given the persistent diarrhoea, fevers and radiographic appearance of metastatic spread, the patient underwent diagnostic workup for a NET. Subsequent labs for NET were all elevated, including chromogranin A at >2500 ng/mL (normal 0–95 ng/mL), neuron-specific enolase at 13.2 µg/L (normal 3.7–8.9 µg/L) and plasma 5-hydroxyindoleacetic acid at 167 ng/mL (normal <22 ng/mL). Empiric octreotide was started with marked improvement in diarrhoea. A retroperitoneal lymph node core-needle biopsy was performed, and staining revealed no evidence of malignancy.

A DOTATATE PET scan was ordered to evaluate for a NET and demonstrated mild–moderate avidity in several lymph nodes, diffuse physiological avidity in the liver and intensely avid osseous lesions in the spine and hip (figure 2). A right inguinal lymph node excisional biopsy was then performed, and staining was again negative for any markers or histological characteristics of malignancy but was consistent with a reactive/inflammatory pattern with intact lymph node architecture.

Figure 2.

DOTATE positron emission tomography/CT scan with multiple moderate–intense areas of avidity. (A) Spinous process of L1 vertebral body (SUVmax 4.6). (B) Anterior aspect of L4 vertebral body (SUVmax 5.7). (C) Paratracheal lymph node (SUVmax 4.1). (D) Right and left femoral lymph nodes (SUVmax 3.8 and 4.3).

A standard PET/CT scan was obtained to further characterise the source of the perceived malignancy. Multiple fluorodeoxyglucose (FDG)-avid lymph nodes, hepatic lesions, as well as intensely FDG-avid non-contiguous osseous lesions were noted. At this time, an excisional right supraclavicular lymph node biopsy was performed on the basis of FDG avidity on the standard PET CT scan. After discussion with our pathologists, staining for AFB was performed, which is not routinely done on lymph node biopsies at this institution. The staining was positive for scattered mycobacteria and several granulomas. Because all lymph node tissue at that point was paraffinised and AFB staining was inconclusive for Mycobacterium tuberculosis (MTB) versus other species of mycobacteria in this relatively immunocompromised host, a biopsy of an intensely FDG-avid L1 osseous lesion was performed. GeneXpert PCR specific for MTB yielded confirmatory results. The patient was subsequently started on rifampin, isoniazid, pyrazinamide, ethambutol and pyridoxine therapy.

Discussion

The breadth of presentations of extrapulmonary TB and the paucibacillary nature of extrapulmonary specimens make the yield of traditional TB studies low. In most reports of pleural tuberculosis, pleural fluid analysis with AFB smear and mycobacterial culture yields positive results in <10% and 12%–70% of cases, respectively.⁷ Similarly, the sensitivity of AFB smear and mycobacterial culture of peritoneal fluid is low at <2% and<20%, respectively.⁸ In our case, the negative ascitic AFB smear and mycobacterial culture were inappropriately interpreted to place TB as less likely on the differential. Interestingly, the sensitivity of mycobacterial culture from ascitic fluid can be increased to >80% if 1 L of ascitic fluid is concentrated by centrifugation and later cultured,⁹ a step also not performed in our analysis. Another potentially useful indirect marker of TB not used in our analysis is adenosine deaminase (ADA), which has a sensitivity of nearly 100% and specificity of >95% in patients without cirrhosis.¹⁰ One study performed in a low TB prevalence area found ADA to be a useful screening tool with a sensitivity of 96%.¹¹

Also inherent in our diagnostic workup was the failure to analyse data from the lens of the developing world. In the radiologic read of the CT scan, the peritoneal thickening was interpreted as carcinomatosis and the multiple liver lesions were interpreted as likely malignant. This bias towards malignancy is likely due to the increased prevalence of cancer versus TB in the USA. Of note, even the incidence of NETs, a relatively rare malignancy, occurs 25% more often than does TB in the USA.¹² However, our patient was born in, and regularly visits, the Dominican Republic where the incidence of TB is more than 20-fold than that in the USA.¹

Imaging specific for NET were strikingly misleading. DOTATATE PET scan in patients in the USA has shown to have a specificity of 90% for NET.¹³ However, activated macrophages have been shown to have strong expression of membrane somatostatin receptors,¹⁴ and scans with radiolabeled somatostatin analogues are currently being evaluated as a diagnostic technique in patients with infectious granulomatous disease (including TB, leishmaniasis and aspergillosis) in the developing world.¹⁵

Laboratory markers of NET can also be misleading. While we interpreted neuron-specific enolase as a marker of NET, it is currently also being evaluated as a marker of TB infection in Asia.¹⁶ Chromogranin A is sensitive for NET, but the specificity can be adversely affected in patients with renal failure.¹⁷ It can also be elevated in critical illness and correlates well with other non-specific markers of disease such as CRP.¹⁸

This case highlights that many of the laboratory tests and imaging modalities that have become the standard in identifying the presence of a NET can be falsely positive in other conditions, especially granulomatous infections such as TB. While these tests are important diagnostic tools in the workup of neuroendocrine malignancies, their specificity can and should be carefully considered in patients from populations at high risk for TB. A tissue diagnosis remains the most important step in identification of malignancy, even in the presence of impressive suggestive findings.

Learning points.

• Sensitivity and specificity for diagnostic tests can change dramatically depending on the patient population being examined.

• Imaging and laboratory markers that have been shown to be sensitive for neuroendocrine tumours may often be positive in patients with infection including tuberculosis.

• Acid fast bacilli smear and culture have extremely poor sensitivity in pleural and ascitic fluid, while adenosine deaminase in fluid is a very sensitive and specific diagnostic tool in the appropriate population.

• Tissue diagnosis remains one of the most important steps in identifying malignancy and, often, in identifying extrapulmonary tuberculosis.