The progression of post-tubercular kyphosis in 61 children who received ambulatory chemotherapy

was studied prospectively. The angles of deformity and kyphosis were measured for each patient at diagnosis, 3, 6, 9, 12 and 18 months later and every year thereafter for 15 years. During the course of the disease signs of instability appeared on the radiographs of some of the children. These were dislocation of the facets, posterior retropulsion of the diseased fragments, lateral translation of the vertebrae in the anteroposterior view and toppling of the superior vertebra. Each sign was allocated one point to create a spinal instability score. The influence on the progression of the deformity of the level of the lesion, the vertebral body loss, the number of segments involved, the angle of deformity before treatment and the spinal instability score was analysed.

The mean angle of deformity at the start of treatment was 35 deg. This increased to 41 deg at 15 years. Progression occurred during the active phase of the disease and again after cure when variations in progression were observed. Type-I progression showed an increase in deformity until growth had ceased. This could occur either continuously (type Ia) or after a lag period of three to five years (type Ib). Type-II progression showed decrease in deformity with growth. This could occur immediately after the active phase (type IIa) or after a lag period of three to five years (type IIb). Type-III progression showed minimal change during either the active or healed phases and was seen only in those with limited disease.

Multiple regression analysis showed that a spinal instability score of more than 2 was a reliable

predictor of patients with an increase of more than 30 in deformity and a final deformity and a final deformity of over 600. Since signs of radiological instability appear early in the disease, they can be reliably used to identify children whose spine is at risk for late progressive collapse. Surgery is advised in these cases.

J Bone Joint Surg [Br] 2001;83-B:954-62.

Received 12 February 2001; Accepted after revision 26 April 2001

Over 30 million people worldwide suffer from overt tuberculosis and more than two million have the active, spinal form.1 The disease can be successfully treated by antituberculous drugs2-4 and attention has now turned to the management of residual deformity.5-7 In the developing world spinal tuberculosis is the main cause of kyphosis; 15% of patients treated conservatively have a considerable increase in kyphotic deformity, which in 3% to 5% is more than 60 deg.5-7 A severe kyphosis is a major cosmetic and psychological disturbance in a growing child and can result in secondary cardiorespiratory problems and late-onset paraplegia. Correction of an established deformity is both difficult and hazardous with a high rate of complications, even in experienced hands.8 Prevention of deformity must be a priority in the treatment of spinal tuberculosis.

Children under the age of ten years at the time of infection are at serious risk of developing severe deformity.5-7,9 The natural history of spinal tuberculosis in children after infection, however, is not clear since most reports have either drawn conclusions at a single point in time2-4,10 or have considered the mean angle of deformity for all children in the study group.2-4,11,12 Both can cause errors. Deformity may improve or deteriorate at various times and a single study will overlook this (Fig. 1). Deterioration also varies between children. The mean angle of deformity will not show the severe increase in deformity which may occur in a few children or the improvement in others. There has been no previous description of the natural history of posttubercular kyphosis in children which specifically documents the progression of the deformity in each child at yearly intervals until growth is complete. This study reports a follow-up of 61 children (63 lesions) over a period of 15 years.

Patients and Methods

These children formed part of a prospective, multicentre, controlled clinical trial of 304 patients with tuberculosis involving the thoracic or lumbar spine to investigate the efficacy of a short course of ambulatory chemotherapy as compared with a short-course of chemotherapy combined with modified Hong Kong surgery, in the management of spinal tuberculosis. Patients were included if they had clinical and radiological evidence of active spinal tuberculosis involving any vertebral body from T1 to S1 inclusive, and if they were available for follow-up for at least five years. They were excluded if they had a neurological deficit, significant extraspinal disease which could affect the management or response to treatment, a history of previous antituberculous therapy for 12 months or more or previous major surgery for spinal tuberculosis.

Of the ambulatory group, 61 patients with 63 lesions were aged 15 years or younger at the time of diagnosis and were selected for this study. There were 26 boys and 35 girls with a mean age at the start of treatment of 6 years and 11 months (2 to 14). Twenty-eight patients (29 lesions) were below the age of five years, 20 (21 lesions) were between six and ten years old and 13 (13 lesions) between 11 and 15 years. Twelve children who had involvement of the T1 to T10 vertebrae were considered to have thoracic lesions, 31 with involvement of T11 to L2 to have thoracolumbar lesions and 18 with lesions below L2 to have lumbar lesions. There were 154 vertebral bodies involved, 38 thoracic, 75 thoracolumbar and 41 lumbar. The mean number of vertebral bodies involved per patient was 2.7 for thoracic lesions, 2.4 for thoracolumbar lesions and 2.2 for lumbar lesions.

For each patient, the angles of kyphosis and deformity were measured at diagnosis and on follow-up radiographs at 1, 3, 6, 9, 12 and 18 months and every year thereafter for 15 years (Fig. 2). In lower lumbar lesions, where LS was affected, the angle of kyphosis was measured by drawing lines along the posterior border of the first normal upper vertebra and the posterior margin of S1. Intra- and interobserver variation in these measurements was analysed in a pilot study in which three observers measured 15 radiographs and repeated the assessment six weeks later. The mean difference was 2.20 and 96% of the differences were less than 5 deg. Using the statistical method of confidence limits14 it was determined that if a measured difference was more than 4 deg, there was 95% level of confidence that this represented a true change.

The vertebral body loss (VBL) at the start of treatment was measured by carefully studying the anteroposterior and lateral radiographs of each patient and dividing each vertebral body into ten equal parts. The amount of destruction in each vertebral body was assessed and then added up to give the total vertebral body loss. At the start of treatment the mean vertebral body loss was 1.6; for thoracic lesions it was 1.8, for thoracolumbar lesions 1.5 and for lumbar lesions 1.4. For patients less than five years of age the vertebral body loss was 1.8; for those aged six to ten years it was 1.7 and for those over 11 years 1.1.

The progression of the angles of deformity and kyphosis was plotted graphically for each patient, which made it possible to view clearly the progression over 15 years and the changes at times of rapid growth. Changes which occurred during the active phase of the disease and after complete cure were documented.

Radiological signs of 'spine at risk'. During the course of the disease, as the severity of collapse increased, the posterior arch became disrupted by dislocation of the facet joints in some children. In severe anterior destruction, facet dislocation led to spinal instability with various radiological features as shown in Figure 3.

Separation of the facet joints. With progressive kyphosis the facet joint at the apex of the curve was subluxed followed by frank dislocation. In patients with severe involvement, there was separation at more than one level with wide distraction of the spinous processes at the corresponding levels.

Posterior retropulsion of the diseased vertebral segments. With progressive destruction, the remnants of the destroyed vertebral bodies were retropulsed. This was assessed by drawing two lines along the posterior surface of the normal vertebra above and below the level of the lesion. Retropulsion was confirmed when bone at the involved level was seen to lie posterior to the drawn lines.

Lateral translation of the vertebral column. Translation was confirmed when the line drawn from the centre of a pedicle of the lower vertebra did not intersect the pedicle of the upper vertebra in an anteroposterior radiograph.