Magnetic Resonance Imaging of Central Nervous System Lesions in HIV/AIDS: About 35 Cases in Libreville (Gabon)

Background: Sub-Saharan Africa is the region most affected by the Human Immunodeficiency Virus (HIV) with an increasing prevalence of related cog-nitive impairments. Magnetic Resonance Imaging (MRI) plays an important role in the early detection of lesions. This work aimed to describe the MRI aspects of different brain lesions occurred in HIV positive patients in our practice. Methods: This was a descriptive cross-sectional study that took place from June 2014 to July 2016 in the medical imaging department of the EL RAPHA private Polyclinic in Libreville, Gabon. It included all patients referred for imaging for the exploration of a Central Nervous System (CNS) lesions at MRI, based on clinical and/or paraclinical arguments. Results: Among the 39 patients included, 19 (48.7%) had a previous brain CT scan, 11 of which were normal (28.2%). Thirty-five (89.74%) patients had a pathological MRI. The main etiologies found were toxoplasmosis (37.14%),


Introduction
Sub-Saharan Africa is the region most affected by the Human Immunodeficiency Virus (HIV) with about 25.5 million cases, or 70% of the global burden of this disease [1]. Statistics also vary between countries in this part of the continent. In the case of Gabon, in 2012, the Demographic and Health Survey found a prevalence of 4.1%. This was twice as high among women, with a rate of 5.8% compared to 2.2% among men [1].
The high vulnerability of organism with multi-visceral damages during this state of untreated immunosuppression by antiretroviral (ARV) drugs is no longer to be demonstrated. Among these, morbid manifestations of the Central Nervous System (CNS) can occur at all stages of HIV infection, particularly in the AIDS disease stage. Their prevalence varies from 40% to 70% according to clinical studies and even reaches 100% in some autopsy series [2].
In Africa, HIV-associated neurological damages vary between 34.0% and 42.5% in Nigeria, and up to 75% in South Africa [3] [4]. They currently affect more than 11 million Africans and thus constitute a diagnostic and therapeutic emergency [4]. The occurrence of neurological clinical manifestations requires an emergency brain scan [5], which is not always possible in our working conditions for many reasons such as the precariousness of patients and the accessibility of advanced diagnostic tools; so much that some of these conditions are over-treated without diagnostic proof.
Magnetic Resonance Imaging (MRI) is much more sensitive than Computer Tomography (CT) for the detection of early and/or small lesions, as well as lesions poorly visualised by CT [6]. Unfortunately, MRI is still not widely available and accessible in most sub-Saharan African countries.
The aim of this work was to show the contribution of MRI in the diagnosis of brain lesions encountered in HIV positive patients in our practice.

Type and Period of Study
This was a cross-sectional descriptive study, with data collection over a 24-month period from July 2014 to July 2016.

Study Site
The study took place in the medical imaging department of the EL RAPHA private Polyclinic in Libreville, Gabon.

Patients
We carried out an exhaustive sampling by systematic census of all HIV-positive patients referred for MRI on suspicion of CNS involvement on the basis of clinical (infectious syndrome, confusional or cerebellar syndrome, or deficiency) and/or biological arguments, and who may or may not have had a previous brain scan.

Data Collection
After collecting epidemiological and clinical information's, an MRI examination was carried out with a high field machine (1.5 Tesla). The protocol included morphological sequences (FLAIR, T1, T2, T2*), diffusion sequences, as well as a 3D T1 Vibe sequence after injection of 10 ml of gadolinium. Two Radiologists carried out the interpretation of the images. In case of discrepancies, the opinion of a third radiologist was required.
All the data was recorded on a survey form filled in by us.

Main Variables
The parameters studied were demographic data (age, sex), clinical signs of orientation, lesions observed on brain CT scan (if applicable), those found on MRI and the diagnoses retained.

Diagnostic Criteria
The diagnosis was retained globally on the basis of a cluster of clinical, biological and radiological arguments:

Statistical Analysis
The data were treated using Epi info 7 software version 2.2.6. The percentages were calculated to describe the qualitative variables. In order to investigate the performance of cerebral tomodensitometry compared to MRI under our practice conditions, the sensitivity, specificity, positive and negative predictive values of the said examination were determined using MRI as reference standard.

Patient Characteristics, Indications and MRI Results
A total of 39 patients were included during the study period.

Performance of Cerebral Tomodensitometry Compared with MRI
Analysis of the results of the 19 patients who had both CT scan and MRI of the brain revealed a sensitivity of 53.33%, a specificity of 100%, a positive predictive value of 100% and a negative predictive value of 36.36% for the CT scan of the brain, using MRI as a reference (Table 3). Specifically, MRI was useful in making a diagnosis in 7 patients who had a non-contributory brain CT scan: 3 cases of HIV encephalitis, 3 cases of PML and 1 case of cerebral toxoplasmosis.

Elemental Lesions Observed on MRI
• Cerebral toxoplasmosis: In the 13 patients diagnosed with cerebral toxoplasmosis, the lesions were multiple in 9 (69%) cases and predominantly left-sided in 9 (69%) cases. 28 lesions were found with cortico-subcortical predominance (23 lesions); the supratentorial site was the most frequent (20 lesions); lenticular lesions were also present (3 lesions) ( Figure 1) as well as those of the thalamus (2 lesions). The enhancement after gadolinium injection was annular in 9 (69%) cases. Subfalcorial involvement was found in 4 (31%) patients. • Tuberculosis: in the 6 patients diagnosed with tuberculosis, there were 2 cases of tuberculoma (Figure 2), 2 cases of cerebral abscess, 1 case of meningoencephalitis and 1 case of meningitis. In the 5 cases of encephalic damage, the lesion was mainly located in the central grey nuclei (3 lesions), followed by the brain stem (1 lesion) and the cerebellum (1 lesion). The lesions were multiple apart from the abscess cases which were rather unique. The latter resulted in diffuse edema with ring enhancement, in contrast to tuberculoma which had a nodular enhancement. In meningitis, a leptomeningeal enhancement after gadolinium injection was found.
• Cortical atrophy: All six patients had an enlargement of the cortical grooves, particularly at the frontotemporal level and dilatation of the lateral ventricles.
• HIV encephalitis: All five patients had diffuse FLAIR and T2 hypersignals of periventricular white matter. The lesions were bilateral and symmetrical, with a T1 isosignal respecting the "U" shaped fibres and the cortical tape. There was no diffusion abnormality, no enhancement after gadolinium injection, and no mass effect (Figure 3). In only one case, cerebellar and peduncular damage were found. • PML: All three patients had hypersignal T2 and FLAIR patches of periventricular white matter, hyposignal T1, without contrast after gadolinium injection. The lesions were bilateral asymmetrical, only supra-tentorial, with damage to the arched fibres and respect for the cortex. The lesions were single or multiple (Figure 4), and there was no mass effect.
• Lymphoma: In one case it was a single right capsulo-lenticular mass, and in the other multiple sus-tentorial masses with periventricular infiltration. These lesions were hyposignal T1, isosignal T2, with intense enhancement after gadolinium injection. Peri-lesional edema and a discrete mass effect ( Figure 5) were associated with these lesions. There was no diffusion restriction.

Discussion
One of the strengths of this preliminary work is that it is one of the few studies that have taken into account imaging aspects, particularly those of MRI in the diagnosis of brain damage during HIV/AIDS in sub-Saharan Africa; the majority of them are more concerned with epidemiological and clinical aspects for logistical reasons. The main shortcoming is the small sample size. It is justified by the financial and geographical accessibility of MRI, which remains very limited in our cities. Despite this insufficiency, it has led to interesting results, which deserve to be considered.
As found in this work, the predominance of toxoplasmosis has also been re- and FLAIR hypersignal infiltration of the corpus callosum and central grey nuclei, with an improvement under anti-toxoplasma treatment [13]. Nodular enhancement was less frequent as reported by Pereira-Chioccola et al. [14].  [11]. The preferred intra-parenchymal locations would be, in order of frequency, the white matter/grey matter junction, the periventricular areas, the posterior cerebral fossa, the central grey nuclei and the mesencephalon [15].

S. M. Kiki et al. Open Journal of Radiology
In this work, they have all been found, even if a predilection for the junction white substance/grey substance on the above tentorial floor has been noted as in the series of Bhanushali et al. and Moufid et al. [10] [16]. In 125 cases of tuberculomas, 60% had a supra-tentorial location, eight of which were multiple and two rare, namely the cavernous sinus and the cerebellopontine angle [16]. Intracranial tuberculoma is usually single and multiple forms are rare [17]. Contrary to Moufid et al. who reported a clear preponderance of single tuberculoma (90%), the cases diagnosed in this work were essentially multiple [16]. Other authors have also reported atypical aspects.
Thus, Koffi et al. described a case of tuberculoma in the form of a plurilobed isosignal lesion poorly limited in T2, heterogeneous with a hypersignal border and a hyposignal background with intense annular contrast after gadolinium injection. The spectrometric analysis was in favour of a glial tumour. The anatomopathological examination allowed the diagnosis to be corrected [18].
In contrast to Costa Gomez et al. who reported 69% of cases of brain atrophy, making it the leading cause of CNS abnormality in HIV-positive people at work [11], this condition was found in only 17.14% of the patients investigated in ours, even though they had low CD4 counts. It has been shown that the older untreated infection is, the greater the loss of volume and cortical thinning in several cerebral territories, with regression after initiation of effective antiretroviral treatment [19]. The reasons for the paradoxically low proportion of brain atrophy in this work are not well known and require further investigation.
Anomalies in the white matter were also found. For example, HIV encephalitis and PML were represented in 14.28% and 8.57% of cases respectively. Elsewhere, they were found in almost similar proportions, ranging from 9.1% to 17.6% for encephalitis, and 5.54% to 14.7% for PML [10] [11].
In these two diseases, MRI has shown a particular performance, allowing the diagnosis of certain patients for whom a brain CT scan was normal. By offering better spatial and temporal resolution than CT scans, which are less sensitive, MRI can provide discriminatory elements to differentiate PML from HIV encephalitis [20]. Indeed, it shows a diffuse FLAIR and T2 hypersignal of the periventricular, bilateral and symmetrical white matter in T1 isosignal in HIV encephalitis, unlike PML where the lesions are unilateral or asymmetrical bilateral, with the participation of "U" shaped fibres in T1 hyposignal then T2 hypersignal and FLAIR. PML lesions can be single or multiple, without mass effect.
It should be noted, however, that although the diagnosis of PML is quite evocative on MRI, some patients may have diagnostic problems related to an unusual presentation or entanglement of pathologies.
As for the location of the lesions, it was only supra-tentorial. In a work published by Toby et al., a localisation at the level of the brain stem and cerebellum was reported [21].
Finally, with regard to lymphoma, a typical case has been diagnosed, according to the data in the literature, with a capsulo-lenticular localization [22]; but Open Journal of Radiology also an atypical case with multiple supra-tentorial masses associated with periventricular infiltration. HIV-infected subjects and the lower performance of CT scans, particularly for some of them. He pleads for the popularisation of this examination, at least in the major sub-Saharan cities, to improve patient care in this region.

Conclusion
MRI has played an important role in the diagnosis of central neurological disorders in HIV-infected individuals. It has been particularly useful in recognising cases of HIV encephalitis, PML and even cerebral toxoplasmosis. Improving its accessibility in sub-Saharan Africa should contribute to better care for people living with HIV. Further work based on MRI, and a larger size, will help to better specify the weight and frequency of these neurological conditions in HIV-infected patients at different stages of the disease evolution.

Ethical Considerations
The publication of the data from this work is in line with the Declaration of Helsinki revised in 2013.