PROXIMAL JUNCTIONAL KYPHOSIS IN ADULT RECONSTRUCTIVE SPINE SURGERY RESULTS FROM INCOMPLETE RESTORATION OF THE LUMBAR LORDOSIS RELATIVE TO THE MAGNITUDE OF THE THORACIC KYPHOSIS

Abstract

Background

Proximal junctional kyphosis (PJK) is defined as: 1) Proximal junction sagittal Cobb angle >≥10°, and 2) Proximal junction sagittal Cobb angle of at least 10° greater than the pre-operative measurement PJK is a common complication which develops in 39% of adults following surgery for spinal deformity. The pathogenesis, risk factors and prevention of this complication are unclear.

Methods

Of 54 consecutive adults treated with spinal deformity surgery (age≥59.3±10.1 years), 19 of 54 (35%) developed PJK. The average follow-up was 26.8months (range 12 - 42). Radiographic parameters were measured at the pre-operative, early postoperative (4-6 weeks), and final follow-up visits. Sagittal alignment was measured by the ratio between the C7-plumbline and the sacral-femoral distance. Binary logistic regression model with predictor variables included: Age, BMI, C7-plumbline, and whether lumbar lordosis, thoracic kyphosis and sacral slope were present

Results

Patients who developed PJK and those without PJK presented with comparable age, BMI, pelvic incidence and sagittal imbalance before surgery. They also presented with comparable sacral slope and lumbar lordosis. The average magnitude of thoracic kyphosis was significantly larger than the lumbar lordosis in the proximal junctional kyphosis group, both at baseline and in the early postoperative period, as represented by [(-lumbar )lordosis - (thoracic kyphosis)]; no- PJK versus PJK; 6.6°±23.2° versus -6.6°±14.2°; p≥0.012. This was not effectively addressed with surgery in the PJK group [(-LL-TK): 6.2°±13.1° vs. -5.2°±9.6°; p≥0.004]. This group also presented with signs of pelvic retroversion with a sacral slope of 29.3°±8.2° pre-operatively that was unchanged after surgery (30.4°±8.5° postoperatively). Logistic regression determined that the magnitude of thoracic kyphosis and sagittal balance (C7-plumbline) was the most important predictor of proximal junctional kyphosis.

Conclusions

Proximal junctional kyphosis developed in those patients where the thoracic kyphosis remained greater in magnitude relative to the lumbar lordosis, and where the sagittal balance seemed corrected, but part of thise correction was secondary to pelvic retroversion.

Level of Evidence

Prognostic case-control study – Level III.

INTRODUCTION

The prevalence of degenerative scoliosis of the lumbar spine has been reported as high as 68% in a healthy population over age 60.¹⁶ The most significant features include loss of lordosis in the lumbar spine, asymmetric disc space collapse, and translational or rotational subluxation. Surgery is recommended for the treatment of intractable low back pain, spinal stenosis with radiculopathy or neurogenic claudication, and for progressive deformity and imbalance. Following neural decompression and stabilization, restoration of sagittal balance may be the most critical factor determining clinical outcomes. ^14,17 Multiple surgical techniques are available to achieve these goals, and include a variety of posterior segmental instrumentations, inter-body spacers, and osteotomies. The proximal extent of the instrumentation is critically determined by choosing the upper instrumented vertebra (UIV) at a segment without posterior column deficiency, without listhesis or rotation, and without junctional kyphosis while also avoiding the apex of a deformity in either the coronal or saggital planes. ^17,18 In spite of these considerations, proximal junctional kyphosis has been described in up to 39% of patients following instrumentations of the lumbar spine for the correction of degenerative deformities, at an average 7.8 year follow-up, ¹⁸ with 59% of the total progression occurring within the first eight weeks after surgery.

Few studies have sought to identify predictors for PJK following adult deformity surgery. ^1,18 The goal of the present study is to identify predictors for this complication, with particular emphasis on the contribution from spinal-pelvic alignment resulting after surgical reconstruction, as measured in standing films at four-to-six weeks postoperatively. Our hypothesis is that those patients in whom the correction of the sagittal balance includes pelvic retroversion are prone to develop proximal junctional kyphosis.

Study Subjects and Methods

Our study design was that of a retrospective case control (Level III) which included 54 patients (45 female, nine male) with an average age of 59.3±10.1 years, who were subject to reconstructive surgery for deformity of the spine. These patients presented with at least one of the following criteria: Coronal Cobb-angle measurement of >30°, lumbar lordosis Cobb angle <30°, thoracic kyphosis Cobb angle >60°, sagittal imbalance (C7-plumbline >5 cm from the sacral endplate), coronal imbalance (C7-P >5 cm from the central sacral vertical line). The clinical records and standing posterior-anterior and lateral scoliosis films were reviewed and measured. IRB approval was obtained.

Radiographic parameters were measured at the preoperative, early postoperative (six weeks), and final follow-up visits (aAverage. 26.8 months: rangel2 - 42 months). Cases (PJK) and controls (no-PJK) were compared by age, sex, BMI, sagittal alignment, and pelvic incidence. Postoperative predictor variables included sagittal balance, lumbar lordosis and thoracic kyphosis Cobb-angle measurement and sacral slope.

Proximal junctional kyphosis was defined according to Glattes et al.¹ by the following criteria: 1) proximal junction sagittal Cobb angle >≥ 10°; and 2) proximal junction sagittal Cobb angle at least 10° greater than the preoperative measurement Our study population was divided into two groups. The a control group (without PJK) consisted of 35 patients who did not develop this complication. A second group consistinged of 19 patients who did develop (PJK) following the index procedure.

Radiographic Measurements

Standing scoliosis postero-anterior and lateral 36-inch films were measured. The positioning of the patients was standardized in 2002 in our Division of Musculoskeletal Radiology, following guidelines recommended by the Spinal Deformity Study Group. ¹³ Sagittal alignment was measured by using the C7-plumbline, which is represented as the horizontal distance in millimeters between a vertical line extending from the center of the C7 vertebrae and the posterior-superior corner of the sacral endplate.

Fixed and postural parameters were measured to describe the anatomic characteristics of the lumbar-pelvic junction. The pelvic incidence corresponds to the angle formed between a perpendicular line to the sacral endplate at its midpoint, and a line projecting from this same point to the center of the femoral heads. If the femoral heads did not completely overlap on the standing lateral film, then the midpoint between the lines connecting the centers of both femoral head was used.

Postural parameters included sacral slope, which is the angle formed between the sacral endplate and a line horizontal to the ground; lumbar lordosis, or the Cobb-angle measurement between T12 and S1 endplates; and thoracic kyphosis, or the Cobb-angle measurement between T1 and T12. Finally, the sacral-femoral distance is the offset between the center of the femoral heads and the posterior superior corner of the sacrum.

The ratio between the C7-plumbline and the sacral femoral distance was used to describe the position of the C7-plumbline with respect to the posterior superior corner of the sacral endplate and the center of the femoral heads. This positional value is relative and is affected by the dimensions of the thoracic kyphosis and lumbar lordosis Cobb-angle measurement, as well as the degree of pelvic retroversion. A C7-plumbline/sacral femoral distance >1 represents a C7-plumbline projected anterior to the center of the femoral heads. If this ratio is equal to 0, then the C7-plumbline falls directly over the sacral endplate, and finally, if the ratio is <0, then the C7-plumbline falls behind the posterior superior corner of the sacral endplate. Finally, if the ratio is in the 0-1 range, this represents a C7-plumbline projecting between the posterior superior corner of the sacral endplate and the center of the femoral heads.

Data Analysis

Patient data was accessed from the University of Iowa electronic medical record. Patients without PJK and those who developed PJK were compared for their baseline demographic characteristics. Non-parametric analysis was performed using the Chi-squared test, and Student’s t-test was used to compare continuous variables. Bonferroni’s correction was applied for multiple comparisons and a p-value of 0.0125 was accepted as statistically significant. Instrumentation types and levels were classified as discrete variables. Binary logistic regression was used to model predictor variables for the development of PJK (dependent variable). Results from uni- and multi-variate analyses are described. Odds ratios and the corresponding 95% confidence intervals are represented. A p-value of 0.05 or less was considered significant.

RESULTS

Of the 54 patients identified during the study period, 19 (35%) developed PJK. The average follow-up was 26.8months (range 12-42months). The average age of the study group was 59.2±10.2 years and the average BMI was 29.1±6.4.

Both the group that developed PJK and the group that did not develop PJK were comparable in age, sex distribution and BMI. Prior to the index surgery, both groups presented with comparable positive sagittal balance (Table 1). Although the average lumbar lordosis was comparable for both groups prior to surgery (29.9±28.2 vs. 34.1.2±15.6; p≥0.494), the average thoracic kyphosis was of greater magnitude in PJK patients (25.9±12.4 vs. 37.3±19.2; p≥0.044). As a result, the average difference between the magnitude of the lumbar lordosis and the thoracic kyphosis was significantly smaller in PJK patients (6.6±23.2 vs. -6.6±14.2; p≥0.012).

TABLE 1.

Comparison of measured values of sagittal alignment between patients who did not develop PJK (no PJK group) and patients who developed PJK (PJK group) at baseline, early post-operative period (six weeks) and final follow-up. Difference between the magnitude of the thoracic kyphosis and the lumbar lordosis Cobb-angle measurement ≥ (-LL - TK); ratio between the C7-P distance to the sacral endplate and the sacral femoral distance - C7-P/SFD; sacral slope - SS; pelvic Incidence ≥ PI. All values are expressed as mean ± SD. A p>0.012 was considered statistically significant.

	NoPJK	PJK	p-value
Baseline
C7-P/SFD	1.8±2.8	1.2±1.0	0.305
ILLI-TK	6.6±23.2	−6.6±14.2	0.012
SS	31.4±16.1	29.6±8.5	0.578
PI	58.4±16.6	56.6±10.0	0.650
6-week Follow-Up
C7-P/SFD	1.4±2.1	0.2±1.3	0.040
IllI-TK	6.2±13.1	−5.2±9.6	0.004
SS	35.7±8.0	30.0±7.9	0.034
Final Follow-Up
C7-P/	1.1±1.4	1.4±0.9	0.515
ILLI-TK	2.6±14.9	−17.5±21.0	0.002

Following reconstructive surgery, the early postoperative radiographs (at six weeks) revealed that PJK patients had an improvement in their lumbar lordosis from -34.1° ±15.6° before surgery to -46.5° ±11.9° six weeks after surgery. This was comparable to the average lumbar lordosis improvement from -29.9° ±28.2° to 44.5° ±11.6° in patients without PJK Nevertheless, the average difference between the magnitude of the lumbar lordosis and the thoracic kyphosis remained significantly smaller in PJK patients (6.2 ±13.1 vs. -5.2 ±9.6; p≥0.004) (Table 1). The average sacral slope increased in patients without PJK (from 31.2° ±15.9° to 35.7° ±8.0°, while it was unchanged in the PJK group (29.3° ±8.2° to 30.4° ±8.5°) (Figure 1). Subsets of patients with and without PJK in which the C7-plumbline had been restored closer to the sacral endplate after surgery (C7-plumbline/sacral femoral distance <1), were compared separately. The sacral slope was significantly higher in this subgroup of patients without PJK (40.3° ±7.7° vs. 30.4° ±8.2°; p≥0.01).

Figure 1.

Example of a patient with severe sagittal imbalance with evidence of pelvic retroversion as well as thoracic lordosis. The postoperative result (six weeks) shows a significant increase in the lumbar lordosis Cobb-angle measurement, and improvement in the C7-plumbline. There is no more evidence of pelvic retroversion, and at final follow-up proximal junctional kyphosis is not observed.

Results from the logistic regression model are displayed in Table 2. The univariate model weighed age, the difference in magnitude between lumbar lordosis and thoracic kyphosis, and the sacral slope as significant predictors of PJK In this model, for every additional year of life the risk of developing PJK increases by 8.5% - for every additional decade, the risk increases 85%. In the multivariate model, the most significant finding is that the greater the difference in magnitude between lumbar lordosis and thoracic kyphosis (odds ratio 0.861,95% CI 0.771-0.961), the lower the risk of developing PJK For every 10° of difference, the risk is decreased by 140%. The position of the C7-pPlumbline was also a significant predictor. This parameter is measured in millimeters, and for every centimeter of increase in the position of the C7-plumbline, the risk of PJK decreased by 30%.

TABLE 2.

Results from binary logistic regression model. The dependent variable, PJK, represents the development of proximal junctional kyphosis. Odds ratio (OR) and 95% confidence intervals (CI) are described for measures at early follow-up (six weeks). The predictor variables include the difference between the magnitude of the thoracic kyphosis and the lumbar lordosis Cobb-angle measurement ≥ (-LL - TK) expressed in degrees, the C7-P distance to the sacral endplate in nun’s, sacral slope ≥ SS in degrees, pelvic incidence ≥ PI in degrees. All values are expressed as mean ± SD. A p<0.012 was considered statistically significant.

	Univariate Model		Multivariate Model
Variable	OR (95% CI)	P value	OR (95% CI)	P value
Age	1.085 (1.015-1.161)	0.0122	NS	Ns
BMI	NS	NS	NS	NS
C7-P	NS	NS	0.972 (0.949-0.996)	0.0055
(−LL – TK)	0.924 (0.865-0.986)	0.01	0.861 (0.771-0.961)	0.0121
PI	NS	NS	NS	NS
SS	0.912 (0.834-0.997)	0.0349	NS	NS

At final follow-up, the C7-plumbline of PJK patients had returned to a position closer to its preoperative location, anterior to the centers of the femoral heads, as demonstrated by the average C7-plumbline/sacral femoral distance ratio of 1.35 ±0.9 (Table 1). Simultaneous to the development of PJK the sacral slope was reduced to values similar to the pre-operative average (Chart 1), and was significantly smaller than that of patients that did not develop PJK (34.1° ±9.7° in no PJK vs. 26.2.° ±10.6° in PJK; p≥0.011).

Chart 1.

Longitudinal profile of measured values of sagittal alignment between patients who did not develop PJK (Nno PJK) and patients that developed PJK (PJK group) at baseline, early postoperative period (six weeks) and final follow-up. Difference between the magnitude of the thoracic kyphosis and the lumbar lordosis Cobb-angle measurement ≥ (-LL - TK); Ratio between the C7-P distance to the sacral endplate and the sacral femoral distance ≥ C7-P/SFD; sacral slope ≥ SS. All values are expressed as mean ± SEM.

The type of instrumentation was classified between screw-only constructs (25/54) and hybrid constructs with the use of proximal hooks (29/54). These types of constructs were used variably but with similar proportions in patients who developed PJK (12/19 hybrid constructs) and those that did not develop PJK (17/35 hybrid constructs) (p≥0.14).

The extent of the instrumentation was divided into three groups: 19/54 patients were fused to the upper thoracic segments (from T2-T6); 22/54 patients were fused to the lower thoracic segments (from T8-T12); and 13/54 patients were fused to the lumbar segments (from only L1 or L2). Of the patients undergoing lumbar instrumentations, including from L1 or L2 level to S1, five of eight (62.5%) developed PJK which contrasts with five of 22 (22.7%) of the patients instrumented from lower thoracic T8 or T10 to S1 (p≥0.014).

DISCUSSION AND CONCLUSIONS

The prevalence of proximal junctional kyphosis was found to be 35% of this study cohort, which is comparable to previously described prevalence of 39% at 7.8-year follow-up in adult patients. ¹⁸ Both studies report a higher prevalence than the 26% of PJK described following surgery for scoliosis in adolescents, at 7.3-year follow-up.²⁰

In the present study, the predictor variables associated with an increased risk of developing PJK included a thoracic kyphosis Cobb-angle measurement exceeding that of the corresponding lumbar lordosis Cobb-angle measurement, and correction of sagittal balance. For every 10° increase in the difference between the magnitude of the lumbar lordosis and the magnitude of the thoracic kyphosis, the risk of developing PJK diminished by 140%. This implies that pre-operative planning should include sufficient correction of the lumbar lordosis, or should additionally address the thoracic kyphosis. Unfortunately, in patients with sagittal imbalance and a decreased lumbar lordosis Cobb-angle measurement, we can frequently observe thoracic hypokyphosis or even lordosis, making it difficult to assess the true magnitude of this curve. Once the lumbar lordosis is corrected, the thoracic kyphosis will correspondingly increase. This has been recently documented,²⁵ and the reciprocal regional alignment changes in the non-instrumented thoracic curve are proportional to the correction of the sagittal lumbar deformity.

The second predictor variable in the multi-variate model was the position of the C7-plumbline. Previous studies concluded that there is a lack of association between postoperative positive sagittal balance and the development of PJK in adults^1,18 as well as in adolescent scoliosis.^19,20 Our results show that in adults, with an average age of 59.2 ±10.2 years, the more positive the sagittal balance is found to be, the less risk there is for development of PJK - there is a 3% decrease for every 1mm in the position of the C7-plumbline. This also means that a 5cm increase in the C7-plumbline would represent a 150% decrease in the chance of developing PJK In a recent review of the literature,²⁴ there was a progressive increase in positive sagittal balance throughout a lifetime, reaching average C7-plumbline values of 40 ±37mm in asymptomatic individuals, with an average age of 76.3 years. Particularly in elderly individuals, correcting the C7-plumbline to close to 0 may represent overcorrection, and will eventually force patients to carry their C7-plumbline forward by developing PJK A likely target goal should be approximately 50mm in front of the posterior corner of the sacral endplate. Additionally, values for sacral slope in these patients (average 30.0 ±7.9°) are significantly lower than those found in normative data. The sacral slope from asymptomatic individuals from different age groups shows a high variability, but this remains relatively unchanged throughout a lifetime, from an average of 41.1 ±8.2° in adolescents to an average of 42 ±9.6° in individuals with an average age of 76.3 years.²⁴ A low sacral slope in the early postoperative period should warn of the persistence of pelvic retroversion (Figure 2), representing incomplete correction of the sagittal alignment, corroborating our hypothesis.

Figure 2.

Diagram representing me mechanism of pelvic retroversion. In a) the C7-plumbline is representing severe sagittal imbalance secondary to degenerative loss of lumbar lordosis. The sacral slope remains at a normal value of 40°. In b), the pelvis is in retroversion, represented by a smaller sacral slope, with a significant increase in the sacral-femoral distance. The goal of transferring the C7-plumbline to the sacral endplate has been achieved, but the hip joint has been placed in a position of forced extension.

In pelvic retroversion, the hips are in a position of maximal extension (Figures 1,2). Additionally, the pelvis is “locked” to the spine by the instrumentation. During the terminal stance phase of gait, under normal conditions, the pelvis presents a transitory increase in its tilt angle which is rapidly followed by 5°-10° of hip extension during heel-rise.²² If the spine is fixed to the pelvis, the C7-plumbline necessarily needs to be thrust into an anterior position to effectively extend the hip during heel raise and toe-off. Repeated cycles of translation of the C7-plumbline likely have an effect on the transition between the rigid instrumentation and the movable spine, resulting in failure of either the bony or ligamentous architecture. Future gait analysis studies are necessary to demonstrate the presence of abnormal kinematics and excessive moments at the transitional segment.

Lordosis in the lumbar spine results mostly from the trapezoidal shape of the lumbar intervertebral discs. The two most caudal intervertebral discs are responsible for 70% of the lumbar lordosis.²³ Particularly in these segments, the degenerative process of the lumbar spine leads to progressive loss of lordosis. As these discs degenerate, the endplates become progressively parallel, resulting in anterior displacement of the C7-plumbline (Figure 2). Several mechanisms are available to correct this loss of sagittal balance, including flattening of the thoracic kyphosis Cobb angle, pelvic retroversion and flexion of the knees while standing.

Pelvic retroversion has been previously described as the forward displacement of the center of the femoral heads (sacral femoral distance), carrying a concomitant change in the sacral slope and pelvic tilt As a result, a progressively horizontal sacral endplate can be observed (Figure 2). This mechanism contributes to the correction of a positive sagittal balance, bringing the C7-plumbline back to the sacral endplate. It is limited by the anterior iliofemoral ligament and the available degree of extension of the hip joints. Normal gait requires approximately 30° of flexion at heel-strike and 10° of extension at toe push-off.²¹ This implies that if the hips are at maximal extension to achieve pelvic retroversion, the trunk must necessarily pitch forward into an unbalanced posture to accomplish a step. In the presence of a spinal-pelvic fusion, the ground reaction forces produced during locomotion are thus transmitted to the junctional region.

Different combinations of compensatory mechanisms for sagittal imbalance are observed in different patients and depend on the degree of flexibility and strength of the involved joints and muscles. For example, a patient with a hip flexion contracture would not be able to use the mechanism of pelvis retroversion. The same is valid for a patient with a thoracic kyphosis that has been stiffened by osteoporotic collapse and advanced spondylosis, who will not be able to provide for thoracic hypokyphosis.

Factors related to the type and extent of the instrumentation have also been studied. Specifically, the use of hook-claw constructs with compressive forces at the upper instrumented vertebra have also been implicated in the development of PJK¹⁵, particularly in adolescents.^19,20 In the present study, all patients were instrumented posteriorly and no differences were found in the proportion of screw-only versus hybrid constructs when comparing patients that did and did not develop PJK. The number of instrumented levels seemed to be a contributing factor for the development of PJK, particularly when comparing instrumentations extending to the thoracic-lumbar junction. In the present series, PJK seemed to be significantly more frequent in patients who were instrumented from L1 or L2 to the sacrum (5/8 cases) as opposed to those who were instrumented from T8 or T10 to the sacrum (5/22 cases). Our preference today is to end these instrumentations in the lower thoracic spine instead of the upper lumbar spine.

Finally, other patient characteristics, such as age and body mass index have been previously evaluated. ¹⁸ In the present study, these variables did contribute in the multivariate analysis and were also comparable for patients with and without PJK.

This study is a retrospective case control study (Level III evidence), with an average follow-up of two years. One limitation of this study is our relatively small group of cases and controls - the multi-factorial nature of this problem warrants further multi-center efforts with greater case volume. With regards to the follow-up period, PJK seems to develop early in both adolescents²⁰ and adults, ¹⁸ with much less significant progression after two years of follow-up. A previously published prevalence study with 7.8 years of follow-up demonstrated that 59% of the cases developed this complication during the first eight weeks after surgery.¹⁸ There seems to be little change between eight weeks and the following two years of follow-up, but more significant changes seem to occur thereafter, which was evidenced at the end of the minimum five-year follow-up (average 7.8 years, range 5-19.8 years).¹⁸

In conclusion, our results should guide surgeons in some of the necessary goals of reconstructive surgery of the spine. The results suggest the most important factor is achieving a lumbar lordosis Cobb angle that exceeds the thoracic kyphosis Cobb angle. This is achieved by either a greater restoration of the lumbar lordosis and/ or by reducing the thoracic kyphosis to a smaller value. Also of importance in elderly patients, the C7-plumbline should be restored to within 5 centimeters of the posterior endplate of the sacrum

Finally, in certain cases, the C7-plumbline may seem adequately restored due to the fact that the patient is using retroversion of the pelvis to compensate for sagittal imbalance. Fusing the lumbosacral spine in retroversion increases the risk of PJK.