Communication between physicians and surgeons interested in spinal deformities depends upon an accurate and universally accepted language. Our language concerning the description and measurement of a curvature in the coronal plane (scoliosis) seems well understood and widely accepted, but that is not true of the description and measurement of a curve in the sagittal plane, especially thoracic kyphosis. PROBLEM No. 1: THORACIC KYPHOSIS, WHAT IS NORMAL? The universally accepted measurement of scoliosis is the Cobb system, which states that a curvature is measured from the end vertebrae of the curve, the end vertebrae being defined as those most tilted from the horizontal on an upright radiograph. The measurement of thoracic kyphosis is confusing, as some authors routinely measure T2-T12, T4-T12, or T5-T12, even though these may not be the maximally tilted vertebrae. Other authors use the maximally tilted vertebrae, whatever these might be. This discrepancy in measurement technique leads to confusion as to what is “normal” and to exactly what is being reported. Inability to accurately define the uppermost maximally tilted vertebra can lead to improper selection of the upper end vertebra for fusion in Scheuermann kyphosis. For example, Propst-Proctor and Bleck1 measured 104 normal children using T5-T12 as the end points in all subjects. The mean kyphosis was 27 degrees, ranging from 21 to 33 degrees. On the other hand, Boseker et al2 measured 121 normal children using the maximally tilted vertebrae (“true kyphosis” technique) and found a mean kyphosis of 33 degrees, ranging from 25 to 42 degrees (1 SD) and 20 to 50 degrees (2 SD). There were 4 children with kyphosis between 15 and 19 degrees and 4 between 51 and 54 degrees, these 8 lying outside the 2 SD range. These authors found the maximally tilted vertebrae to be T2-T12 in 49%, T2-L1 in 13%, T2-T11 in 10%, T1-T12 in 10%, T3-T12 in 7%, and “other” in 11%. This is the only study of which we are aware that accurately defined the various end vertebrae (Fig. 1).FIGURE 1.: The data from the article by Boseker et al2 shows a classic bell-curve distribution.Takemitsu et al3 measured 519 Japanese children, 253 boys and 266 girls, using the maximally tilted end vertebrae (true kyphosis) technique. They found similar measurements in girls and boys and noted a small increase in the mean kyphosis during growth. Boys increased from a mean of 35 degrees (±10 degrees for 1 SD) at ages 6 to 11 years to a mean of 41 degrees (±11 degrees for 1 SD) at ages 13 to 16 years. Girls increased from a mean 29 degrees to a mean 41 degrees for the same time periods. These large ranges of normal were emphasized in the French study of adults by Stagnara et al.4 The mean kyphosis in their study was 37 degrees, but with a range from 7 to 63 degrees. In their conclusion, they recommended avoiding the words normal or “abnormal.” Bernhardt and Bridwell5 studied 102 normal children, 55 females and 47 males. They noted that thoracic kyphosis usually began at T1 or T2 and found the mean kyphosis to be 36 degrees, ranging from 9 to 53 degrees (±10 degrees for 1 SD). Mac-Thiong et al6 measured 180 children aged 4 to 18 years using the T2-T12 vertebrae. The mean thoracic kyphosis was 43 degrees (±10.4 degrees for 1 SD), range 34 to 54 degrees. In summary of the above studies, when the kyphosis is measured from the upper thoracic vertebrae (not the T5-T12 kyphosis), the mean thoracic kyphosis in children and adolescents ranged from 332 to 43 degrees,6 with very large ranges and SDs. As is shown in Figure 1, these are classic bell-curve distributions where the extremes are difficult to define. Certainly anything from 15 to 50 degrees can be considered normal, and even 15 to 55 degrees could be justified. The old statement found in many texts, and as recently as a 2004 Instructional Course Lecture by Betz,7 that normal is from 20 to 40 degrees is not justifiable. Adding to the confusion was a commercial monograph “Radiographic Measurement Manual,” published by Medtronic Corp and authored by O'Brien et al,8 wherein it states (p. 87) under Adult Deformity “thoracic kyphosis is measured from the cephalad endplate of T2 to the caudal endplate of T12.” However, later on the same page, it states “the maximal measured kyphosis is measured from the upper endplate of the most cephalad vertebra within the kyphotic curve to the lower endplate of the most caudal vertebra.” On page 65, under “Adolescent Idiopathic Scoliosis,” it states “thoracic kyphosis is measured from the upper (cranial) endplate of T2 to the lower (caudal) endplate of T12 using the Cobb method.” From the above discussion, it is seen that this measurement may be measuring the true kyphosis in half the cases. PROBLEM No. 2: HOW DO WE TAKE A GOOD RADIOGRAPH? Why has the T5-T12 area been used in some studies? It would seem that the chief reason was the difficulty in visualizing the T1-T4 area on chest radiographs9 or older spine radiographs. With careful attention to patient positioning and radiologic techniques, decent visualization of the upper thoracic spine can be achieved. When one takes a lateral standing radiograph with the arms at the side, the humeral shafts can overlie the spine, preventing good measurement. To get around this issue, the arms need to be positioned forward out of the way. When a subject simply lifts the arms forward from the body, it induces lumbar hyperlordosis and distorts overall sagittal alignment. To avoid this, 2 techniques have been developed. In the paper by Boseker et al,2 the authors describe testing different positions and finding that the best technique was to have the patient standing in a relaxed posture with the arms forward 90 degrees to the torso and resting on a pole or ladder in a relaxed posture (Fig. 2). Some radiographs are somewhat indistinct in the T1-T4 area and the endplates are not adequately seen for a good measurement. This problem can often be overcome by “recreating” the contour by drawing a line along the anterior and posterior vertebral body cortices, the “best fit line.” Once this has been carried out, then perpendiculars to that line can be used to measure the kyphosis (unpublished data by F. Takeuchi and F. Denis) (Figs. 3, 4).FIGURE 2.: A lateral standing radiograph of a 33-year-old woman with Scheuermann disease. Note that all vertebrae in the upper thoracic spine can be well seen. The arms are forward, resting on a support. Her true kyphosis is 75 degrees and corrected to 48 degrees on a supine hyperextension film.FIGURE 3.: A lateral standing radiograph in which the upper thoracic vertebral endplates are not well seen, but the kyphosis can be measured by reconstructing the spinal contours using the anterior and/or posterior vertebral body cortices (Tekeuchi and Denis technique).FIGURE 4.: A lateral standing radiograph of an adolescent boy with a high thoracic kyphosis secondary to a laminectomy for a spinal cord tumor. Note that with good technique and centering the beam more proximally, excellent visualization of both the thoracic kyphosis and cervical lordosis is seen.Vedantam et al10,11 also tested various arm positions and recommended a 30-degree unsupported forward arm position, noting that a 90-degree forward unsupported arm position created lumbar hyperlordosis. They reported a mean thoracic kyphosis as measured from T3-T12 of 38 degrees. More recently Faro et al12 and Horton et al13 described the “clavicular” position, wherein the patient stands with the arms crossed over the chest and the fists against the opposite clavicle. Both the Boseker/Moe technique and the clavicular technique give a good view of the upper thoracic spine and avoid lumbar hyperlordosis. PROBLEM No. 3: HOW DO WE EVALUATE CORRECTION? When dealing with scoliosis correction, we typically describe the percentage correction on the basis of the pretreatment curve value against the follow-up value (pre−post/pre×100). When a 60-degree scoliosis is corrected to 30 degrees, there is a 50% correction. This is all based on the normal spine having a 0 lateral curvature. A scoliosis correction from 60 to 0 degrees would be a 100% correction. Kyphosis is a totally different problem. As a 0 degree final treatment would be a significantly pathologic thoracic flatback, we need to avoid a percentage correction mentality. As the goal of treatment, whether by bracing or surgery, is to achieve a normal sagittal alignment, we need to express our results in those terms, that is, did we achieve or not a significant number of degrees of correction, and did the patient end up with a normal thoracic kyphosis. Thus, if a patient with Scheuermann kyphosis of 80 degrees has corrective surgery and ends up at 40 degrees, we should not state there has been a 50% correction, but rather that a “full” correction to a normal status has been achieved. We want to know both the number of degrees of correction achieved and whether the patient ended up in the normal range, but the percentage correction is not relevant. With the controversy as to what is the normal range of thoracic kyphosis (15 to 50 degrees or 15 to 55 degrees), it may be best to only state the degrees of correction and not use the term “normal kyphosis.” CONCLUSIONS When we wish to document the true thoracic kyphosis in a subject, the optimal radiograph is taken in the standing position with the arms forward at 90 degrees and resting on a support or in the clavicular position. The kyphosis can then be measured from the uppermost tilted vertebra to the lowermost tilted vertebra, whatever these may be. When we report our results of treatment, we must avoid using “percentage correction” like we do for scoliosis. We need to know the number of degrees of correction and the final number of degrees of the kyphosis. We might also state whether or not the kyphosis was corrected into the normal range, but that is fraught with difficulty considering the huge range of normal.
No takes yet. Share an insight, caveat, or question.
Winter et al. (2009) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: