Nontraumatic Myelopathy in Malawi: A Prospective Study in an Area with High HIV Prevalence

Laboratory procedures at QECH.

If indicated, an HIV test was obtained. Full blood and CD4 counts were carried out. Urine was tested for protein, glucose, and cells by dipstick rapid diagnostic tests.

Lumbar puncture was performed if clinically indicated, and routine analysis of cerebrospinal fluid (CSF) included protein and glucose levels; total leukocyte count and differentiation; and staining with Gram stain, Ziehl–Neelsen stain, and India ink.

Schistosomiasis was diagnosed by demonstration of Schistosoma eggs by microscopic examination of concentrated samples of urine (filtration) and stool (Kato–Katz method). Where possible, three consecutive stool samples and three consecutive urine samples were collected and examined per patient. Microscopy findings were scored as positive or negative; the number of eggs found was not recorded.

Although other helminthic infections were occasionally detected, these were not included in this report.

Radiological imaging.

This included a chest or spinal X-ray and an ultrasound. The radiological examinations were not standardized but were at the discretion of the attending clinician who was guided by the clinical presentation, that is, the symptoms and signs at presentation. MRI scan services were restricted, and an MRI scan was only performed when clinically indicated. The MRI scans were reported by a radiologist (SK).

Surgical intervention.

If indicated, the spinal lesion was surgically explored for drainage or removal of a mass. In some cases, specimens could be examined by a pathologist for histological diagnosis.

Additional laboratory procedures.

Additional examinations were retrospectively performed in stored samples of serum, urine, stool, and CSF by the Erasmus University Medical Center, Department of Virology and Department of Medical Microbiology and Infectious Diseases, and the Leiden University Medical Center, Department of Parasitology, in the Netherlands.

The presence of DNA of Schistosoma species in frozen-stored urine, stool, and CSF samples was performed by real-time polymerase chain reaction (PCR) as described for stool samples before.¹² Only one stool and one urine sample per patient were analyzed by this PCR. Antibodies against Schistosoma species were determined as described before with minor modifications.¹³ The Schistosoma egg antigen ELISA test result was considered negative if < 100 and equivocal if between 100 and 200.

Intrathecal production of Schistosoma-specific antibodies was investigated as described before.¹⁴ The albumin index (albumin in CSF/albumin in serum) was used as a measure for integrity for the blood–brain barrier (BBB). If the albumin index was < 0.0090, the BBB was assumed intact and intrathecal anti-Schistosoma IgG production could be examined by calculating the anti-egg versus anti-worm antibody index, which is defined as follows:

• $\raisebox{1ex}{$\left({{\mathrm{E}}_{\mathrm{c}}/\mathrm{W}}_{\mathrm{c}}\right)$}\!\left/ \!\raisebox{-1ex}{$\left({\mathrm{E}}_{\mathrm{s}}/{\mathrm{W}}_{\mathrm{s}}\right)$}\right.$

If > 4, the index indicates intrathecal schistosomiasis; if 2–4, the index is equivocal; and if < 2, the index is negative.¹⁴

Borreliosis was diagnosed by quantitative determination of specific IgG in serum and CSF against recombinant antigens of Borrelia burgdorferi, using a commercially available chemiluminescence immunoassay (DiaSorin, Saluggia, Italy) that was performed according to the manufacturer’s instructions.

Syphilis was diagnosed by a commercially available Treponema pallidum particle agglutination assay (TPPA) (Serodia, Fujirebio Inc., Tokyo, Japan). Serum and CSF were examined for the presence of specific antibodies against T. pallidum by serial dilution according to the manufacturers’ instructions.

Tuberculosis was diagnosed according to the guidelines of the Malawi Tuberculosis Control Programme and included examination of sputum, lymph node aspirate, or pleural fluid specimens with Ziehl–Neelsen stain and radiological imaging. In addition, the presence of Mycobacterium tuberculosis in CSF was examined by a commercially available PCR assay according to the manufacturer’s protocol (Hain Lifescience, Nehren, Germany).

Molecular virology screening.

All samples were screened for the presence of viral pathogens by real-time (reverse-transcriptase) PCRs with primers and probes used in a routine setting of Erasmus University Medical Center molecular viral diagnostics.¹⁵

For detection of HSV-1, HSV-2, CMV, EBV, VZV, HHV-6, HHV-8, Enterovirus, Parechovirus, and JC virus in CSF samples, a semiautomatic workflow was including a Xiril-α (pre-extraction sample handling), MagNa Pure96 (nucleic acid extraction), Xiril-γ (PCR-setup), LC480-II, and middleware software AuroraFlow (Roche Diagnostics, Almere, the Netherlands). Total nucleic acids were isolated using an input and output volume of 100 μL and 300 μL, respectively. For the detection of chikungunya virus, HIV-1, HIV-2, hepatitis B virus (HBV), and HTLV-1 in serum, nucleic acids were extracted using MagNa PureLC (Roche Diagnostics, Almere, the Netherlands), with the High Pure 200 extraction protocol and the total nucleic acid isolation kit, with an input and output volume of 100 μL and 200 μL, respectively. Because of the limited amount of sample volume, the samples had a final dilution of 6–60× compared with routinely analyzed samples in the diagnostic setting, depending on the available sample volume.

Input volumes for real time reverse transcription polymerase chain reaction (RT-PCR) assays for RNA viruses were 8 µL for entero- and parechoviruses and 20 µL for HIV-1, HIV-2, and chikungunya virus. Input volumes for real-time PCR assays for DNA viruses were 8 µL (for EBV, CMV, VZV, HHV-6, HHV-8, and JC), 10 µL (for HSV-1 and HSV-2), or 20 µL (for HBV). The extracted RNA or DNA was amplified as described previously.¹⁵

For HTLV-1 infection, 5 µL of RNA was used. The extracted RNA was converted to cDNA and amplified essentially as described, using primers P-tRNA (gamma F) (5'-CADKTGGGGGCTCGTCCGGGAT-3) and POL-3 (delta R) (5′-GGCCTGGAGGCGYTCHRGTTTAAMGG-3′).¹⁶ The PCR temperature profile was 7 minutes at 95°C, 40 cycles of 40 seconds at 95°C, 1 minutes at 65°C and 2 minutes at 72°C, and final extension 10 minutes at 72°C.

A subset of samples was analyzed previously using a metagenomics approach in a virus discovery study.¹⁷

Clinical and final diagnosis.

Based on the clinical assessment during admission, a most likely clinical diagnosis was made, and medical or surgical treatment was commenced as appropriate. After reviewing of the additional laboratory results, a most likely final (differential) diagnosis was made in retrospect.

Neuroschistosomiasis was diagnosed using a positive PCR test result in the CSF or on evidence of intrathecal anti-schistosomal antibody production. Neurosyphilis was diagnosed by using TPPA in serum and in CSF, with an intact BBB. Neuroborreliosis was diagnosed if evidence was found for intrathecal production of anti-Borrelia IgG with an intact BBB. Myelopathy associated with HSV-1, HSV-2, CMV, EBV, VZV, HHV-6, HHV-8, Enterovirus, Parechovirus, JC virus, and HTLV-1 was diagnosed on the presence of a positive PCR result in the CSF. M. tuberculosis infection of the central nervous system was confirmed by CSF analysis including Ziehl–Neelsen staining and PCR. Spinal tuberculosis was considered in case of a positive Ziehl–Neelsen stain of the sputum, and/or X-rays of the chest and spine, in combination with an MRI suggestive of tuberculosis, or a total leukocyte count and differential cell count, and protein and glucose levels in the CSF suggestive of tuberculosis. Tumors or metastases were either diagnosed by histological examination of a biopsy or material obtained during surgery.

Data collection and analysis.

The data were entered on a patient record form and subsequently analyzed using SPSS 24. Standard tabulation and descriptive methods, such as means, were used for descriptive purposes. Means were compared using independent sample t-test and proportions by using chi-squared test. A significance level of P < 0.05 was used. Logistic regression was used to explore whether combinations of variables were associated with common final diagnoses.

Ethical approval and informed consent.

The study received approval from the College of Medicine Research and Ethics Committee and College of Medicine, Blantyre, Malawi. Informed consent was obtained from all patients.

Detection of schistosomiasis.

In 57 of the 58 included patients, an attempt to diagnose schistosomiasis could be made by either microscopy or PCR analysis of urine (n = 57, 98%) or stool (n = 47, 81%).

Three consecutive urine samples could be examined by microscopy in 46 patients (79%). For stool, this was in 24 patients (41%). Schistosoma eggs were seen in seven patients (12%). In six patients, eggs were demonstrated in stool: in five cases, Schistosoma mansoni was found, and in one case, Schistosoma haematobium was found. In three patients, S. haematobium eggs were demonstrated in urine. In two patients, eggs were found both in stool and urine samples; one had double infection with S. mansoni and S. haematobium, and one had S. haematobium eggs in both sample types. The PCR analysis demonstrated Schistosoma spp. DNA in another seven patients (n = 2 in stool and n = 5 in urine). The presence of Schistosoma eggs observed in three patients (n = 2 in stool and n = 1 in urine) could not be confirmed by PCR. In total, 14 of the 57 examined patients (25%) were microscopy and/or PCR positive, so these patients were considered to have active schistosomiasis.

No DNA of Schistosoma spp. could be detected in any of 39 (67%) tested CSF samples.

Of 52 serum samples examined, 13 patients (25%) were found positive for antibodies against Schistosoma worm or egg antigens; 34 patients (65%) were negative, and in five patients (10%), the result was equivocal. In the CSF of 17 patients examined, using the egg-worm index, two patients had a score of > 4, whereas in another two patients, the result was equivocal (score 2–4). Hence, neuroschistosomiasis was included in the final differential diagnosis in these four cases (Supplemental Table 1).

Detection of virus infections.

Of 55 serum samples tested for HIV at presentation in QECH, using routine screening with two rapid antibody tests, 23 (40%) were positive, with a mean CD4 count of 221 cells/mm³ (range, 39–535); in 67%, the CD4 count was < 200 cells/mm.³

The additional laboratory investigations retrospectively performed in the Netherlands were carried out on 50 serum samples and 35 CSF samples. Using routine real-time RT-PCR assays, HIV-1 was detected in 11 of 50 (22%) of serum samples and in nine of 35 CSF samples (26%); all but one had a CD4 count < 200 mm³. In eight patients, these findings correlated in serum and CSF; in one patient, there was lack of serum. In seven of nine who were positive for HIV in CSF, HIV-1 was the only virus detected; in one HIV-1–positive patient, EBV also was demonstrated in CSF as well as HBV in serum. Another HIV-1–positive patient was also positive for human cyclovirus.

Cyclovirus was detected using a metagenomics approach in 15% and 10% of serum samples, and CSF samples, respectively.¹⁷

Hepatitis B virus was detected in the serum of six patients, two of whom had hepatocellular carcinoma. In one HIV-1–negative patient, HHV-6 was detected in CSF.

No evidence was found for VZV, CMV, HSV-1, HSV-2, enterovirus (including polio), parechovirus, JCV, or HTLV-1 infections (only CSF samples examined) or HIV-2 and chikungunya virus (both CSF and serum samples examined) (Supplemental Table 1).

Detection of bacterial infections.

No positive Ziehl–Neelsen stain was found in the CSF samples examined.

All 41 examined CSF samples tested for mycobacterial infection by PCR were negative.

Antibodies against T. pallidum were detected by using TPPA in eight of 53 (15%) examined serum samples. In four of those patients, the CSF was enriched for T. pallidum–specific IgG, suggesting neurosyphilis that was, therefore, considered to be in the differential diagnosis in these four cases (Supplemental Table 1).

None of the 53 examined CSF samples were enriched in Borrelia-specific IgG.

Table 2 shows the distribution of symptoms, ASIA score, and main laboratory findings at presentation according to the most likely final differential diagnosis.

Using multivariate logistic regression, HIV status was found predictive of the diagnosis of spinal tuberculosis (P = 0.018). The total white cell count (TWC) and ASIA score were not predictive of this diagnosis.

In Table 3, the most likely final diagnoses are summarized in categories according to HIV status. Of 58 patients studied, myelopathy was diagnosed in 46 (79%), whereas in others, radiculopathy by disk herniation (two patients, 3%) or polyneuropathy (4 patients, 7%) was considered likely. In four patients (12%), the diagnosis remained entirely unknown, whereas in two patients, the most likely primary differential diagnosis was not established.

Limitations of the study.

This study followed the current clinical practice regarding diagnosis and management of patients presenting with non-traumatic myelopathy. Given the limited resources a practical approach was often followed, and a confirmed histological diagnosis was only made in a minority of patients, restricted to those in whom surgery was indicated. MRI imaging is not unrestricted in our setting but contributed considerably in those in whom the diagnosis was largely unclear. Invasive diagnostic procedures aimed at spinal lesions carry risk and this should be balanced against a thorough clinical assessment including imaging and subsequent therapeutic approach.

This study shows the predominance of tuberculosis as the cause of nontraumatic myelopathy, largely driven by HIV infection. In most cases, spinal tuberculosis can already be expected on evidence of tuberculosis found elsewhere (sputum examination, chest X-ray, and lymph node aspiration). The MRI imaging may provide strong support for tuberculosis. Recently, the use of the Xpert MTB/RIF assay in urine was reported as an aid to the diagnosis of tuberculous spondylitis.²⁶ Starting anti-tuberculous treatment based on these findings seems justified in the absence of other diagnostic possibilities. In patients with a mass that compresses the cord, surgical intervention is necessary. This shows that paraplegia patients should be assessed in a multidisciplinary team by internists or pediatricians, respectively, and orthopedic surgeons to determine the best approach to diagnosis and management.

Although only a few potential cases of SCS were identified, empirical treatment with praziquantel seems still justified in high prevalence areas, as treatment is safe and cheap.

Paraplegia is a devastating neglected (tropical) syndrome for which a minimum of diagnostic tools including MRI scanning should be available to limit prolonged and irreversible morbidity as the result of misdiagnosis, exposure to inappropriate drugs, and invasive procedures.