From December 17^th 2007 to 31^st December 2009, consecutive HIV-TB co-infected patients, qualifying for ART according to National ART treatment guidelines, were prospectively enrolled in a cohort to study the incidence of the immune reconstitution inflammatory syndrome at the National Tuberculosis and Leprosy Programme clinic [8] .

Study eligibility criteria included adults (>18 yrs), living within a 20 kilometre radius of the hospital, with a positive TB diagnosis as defined by microbiology or by the WHO criteria [9] , starting TB treatment or having taken TB treatment for less than two months and were eligible for ART initiation according to the Uganda national ART treatment guidelines (CD4+ T-cell counts ≤ 250 cells/µl) [10] . Exclusion criteria were liver function abnormalities (alanine and aspartate transaminases greater than five times the upper limit of normal) and renal failure.

All patients were subjected to a detailed clinical history, physical examination, and laboratory investigations. Other investigations were performed based on clinical indications and their CD4 counts.

Patient characteristics (age, gender), haemoglobin, CD4 counts measurement were obtained at study enrolment and relevant specimens (sputum, lymph node aspirate, pleural effusion, etc.) were examined by Ziehl Nielsen (ZN) stain and Lowenstein Jensen (LJ) culture for TB (details of the microbiological findings have been published elsewhere) [11] .

A chest X-ray and an abdominal ultrasound were performed at enrolment. An abdominal ultrasound was conducted for evaluation of the abdomen. Two radiologists (HNK, MGK), certified by the National Medical Association Board and each with more than ten years of experience, performed the abdominal ultrasound scans with the patient in supine position. Each sonogram was obtained using a sonography unit (Philips HD7, China), with a broad band (2 - 5 MHz) curve-linear and high a frequency (7.5 - 10 MHz) linear transducer.

A complete abdominal sonographic examination was performed included a detailed evaluation of the liver, gallbladder, bile ducts, spleen, pancreas, kidneys, mesentery, bowel loops, aorta and inferior vena cava. Findings recorded included organ enlargement, altered echogenicity and echo-texture, presence of focal masses (single or multiple), fluid collections, lymphadenopathy, hollow visceral thickening, gut wall matting and vascular assessment. Color flow Doppler was used to further analyze an area or lesion of interest in case of need. After all organs and their contiguous structures were individually examined, a diagnosis was formulated. If no pathology was found a limited number of standard hard copy images were recorded. All sono-images were stored electronically.

Lymphadenopathy: nodes larger than 1 cm in diameter [12] [13] splenomegaly: spleen more than 13 cm in its maximum length [13] [14] , hepatomegaly: longitudinal measurement of the liver at the mid-clavicular line measures more than 15 cm [15] , Centri-lobular/“starry sky” pattern of the liver: decreased echogenicity of the liver with apparent increase in the number of portal vein walls with more than 10 such vessels per image [15] -[17] . The fatty-fibrotic/bright liver pattern: increased echogenicity of liver parenchyma and sound attenuation with decreased definition of portal vein walls [16] [18] . Normal gallbladder wall: ≤3 mm [19] .

Medians and interquartile range (IQR) were calculated for continuous variables. The chi-squared test or the Fisher Exact test was used to compare proportions of patients with and without sonographic abnormalities as indicated. The analysis was performed using STATA 12 software (Stata, East College Station, TX).

The University Research and Ethics Committee, the Committee on Human Research, the Hospital Institutional Review Board and the National Council for Science and Technology approved the protocol. Informed consent was obtained for screening and for enrolment into the study.

Abdominal TB accounts for 15% - 20% of all TB cases in Africa [20] . Patients with abdominal TB tend to present with abdominal pain, weight loss and fevers [5] [7] [11] [21] -[24] . However, a definitive diagnosis in developing countries with high prevalence of HIV is often a challenge due to but not limited to: low sensitivity and specificity of the clinical symptoms, the paucity of radiological and diagnostic modalities, and difficulties in accessing the affected tissues for a diagnostic specimen [6] [7] [23] [25] -[27] .

A number of studies have found that a combination of clinical features and sonographic characteristics may

^aOut of 181 patient with available data.

^†Biliary abnormalities included cholangiopathies, bile duct thickening and or dilatation (n = 3), cholelithiasis with or without cholecystitis (n = 12), ^*Pancreatic abnormalities: I case of chronic calcific pancreatitis with a pancreatic pseudocyst and one case of acute pancreatitis, ^βRenal abnormalities: 26 cases of nephropathy (diffuse enlarged kidneys with diffuse increased echogenicity), 2 cases of cortical renal cysts, 1 case of obstructive hydronephrosis due to a pelvic abscess and I case focal renal masses.

^*Stands for the number of patients with lymph nodes, some patients had lymph nodes occurring in more than one site.

help to diagnose abdominal TB early and therefore lead to earlier treatment and improved survival [5] -[7] [21] [22] [24] [26] -[28] . In resource constrained countries, abdominal sonography is traditionally performed very selectively to evaluate specific abnormalities discovered at clinical examination or when the patient has abdominal complaints. However, the signs and symptoms of the disease may be masked by concurrent illness and weak immune response, reducing the sensitivity of relying on the tradition clinical assessment [24] . In addition, abnormal ultrasound findings have been documented in this and other previous studies in patients with HIV-TB co- infection without abdominal symptoms, especially in patients with severe immune deficiency [6] [7] [24] [26] [27] .

Therefore a screening abdominal ultrasound scan especially in severely immunosuppressed HIV-TB co-in- fected patients, would be useful before initiation of ART to detect pathology that would be missed just by physic- cal examination or by performing ultrasound only in selected symptomatic patients. Current advances in ultrasound technology and scan heads allow greater spatial and contrast resolution, for early detection of comorbidities among patients with HIV-TB co-infection even before they become clinically apparent [6] [27] [28] .

In addition, the decrease in prices for ultrasound equipment makes this imaging modality easily available and a cost effective screening and diagnostic tool especially in resource constrained countries.

Abdominal sonographic findings of TB in patients with advanced immune deficiency include: mesenteric hypo-echoic lymphadenopathy [21] [22] [28] -[31] ascites [21] [26] [27] , multiple splenic hypo-echoic nodules and thickened matted bowel loops. Some studies however have found ascites to be more common in HIV negative patients [6] [7] [32] . Evidence of TB elsewhere, like in this study may provide circumstantial evidence in favor of a diagnosis.

Multiple lymphadenopathy, the commonest finding observed in 38% of the studied patients is much lower than the 55.3% - 96.7% reported in studies that only included patients with extra-pulmonary TB. These tend to be poorly vascularised (Figure 1(a)) and undergo caseation (Figure 1(b)). The lack of retroperitoneal lymphadenopathy despite extensive mesenteric and peritoneal involvement in TB may be explained by the fact that mesenteric lymph nodes drain direct into thoracic duct via the cistern chili [22] .

The presence of ascites should raise the suspicion of TB. Different sonographic patterns of ascites in abdominal TB have been described including free or loculated, with or without stranding/septations like violin strings or lattice appearance (Figure 2) [26] [27] [33] -[35] .

Variable incidences of hepatomegaly and splenomegaly in HIV-TB co-infection have been described ranging from 19% - 85% [6] [7] [26] [27] . The most common encountered liver abnormality is a diffuse fatty-fibrotic pattern, often due to fatty degeneration of the liver. A study of abdominal ultrasound in adults with AIDS in Zambia and Zaire found 14 cases of TB out of 42 ultrasound guided liver biopsies among patients with focal parenchymal lesions [6] . TB lesions in the spleen may present as multiple small hypo-echoic nodules of 5 - 10 mm, best demonstrated by using the high-frequency probe for better resolution (Figure 3) [21] [26] [27] [30] [36] .

We observed bowel thickening with loss of architecture and definition of the five layers (Figure 4(a)), depicting the “pseudokidney appearance” (Figure 4(b)), but with no increased blood flow. Other studies record bowel

thickening and matting, obstruction and fistula formation [26] [27] [29] [32] [37] . Mesenteric thickening of 15 mm or more, has been attributed to TB [22] .

In the kidneys, Mycobacterium tuberculosis produces multiple hypo-echoic small abscesses, calcified foci and hydronephrosis due to obstruction by retroperitoneal masses [6] [26] [27] [29] [33] . All these changes tend to resolve with TB therapy.

The main limitation of our study is the lack of microbiological or histological confirmation of abdominal TB. We used circumstantial evidence of tuberculosis elsewhere in favor of diagnosis. We can thus not be certain that all the observed abnormal son-ographic findings were caused by TB.