Distal Junctional Failure: A Feared Complication of Multilevel Posterior Spinal Fusions

Objectives: Distal junctional failure (DJF) is less commonly described than proximal junctional failure following posterior spinal fusion, and particularly adult spinal deformity (ASD) surgery. We describe a case series of patients with DJF, taking into account sagittal spinopelvic alignment, and suggest potential risk factors in light of the current literature. Methods: We performed a single-center, retrospective review of posterior spinal fusion patients with DJF who underwent subsequent revision surgery between June 2009 and January 2019. Demographics and surgical details were collected. Radiographical measurements included the following: preoperative and postoperative sagittal and coronal alignment for each index or revision surgery. The upper-instrumented vertebra (UIV), lower instrumented vertebra (LIV), and fusion length were recorded. Results: Nineteen cases (64.7 ± 13.5 years, 12 women, seven men) were included. The mean follow-up was 4.7 ± 2.4 years. The number of instrumented levels was 6.79 ± 2.97. Among the patients, 84.2% (n = 16) presented at least one known DJF risk factor. LIV was frequently L5 (n = 10) or S1 (n = 2). Six patients had an initial circumferential fusion at the distal end. Initial DJFs were vertebral fracture distal to the fusion (n = 5), screw pull-out (n = 9), spinal stenosis (n = 4), instability (n = 4), and one early DJK. The distal mechanical complications after a first revision included screw pull-out (n = 4), screw fracture (n = 3), non-union (n = 2), and an iatrogenic spondylolisthesis. Conclusions: In this case series, insufficient sagittal balance restoration, female gender, osteoporosis, L5 or S1 LIV in long constructs were associated with DJF. Restoring spinal balance and circumferentially fusing the base of constructs represent key steps to maintain correction and prevent revisions.


Introduction
The treatment of adult spinal deformity (ASD) is known for the frequent necessity to perform complex surgeries that usually require long segment fusions to correct and stabilize the deformed spine.These procedures tend to be more and more frequent due to the growing demand of spinal surgery.Aging populations with an increased life expectancy strongly desire a higher quality of life.Older patients pose several challenges and complications that are typically not present in pediatric or adolescent spinal deformity surgery.Numerous medical and surgical complications occur during or after these procedures.Yet, the mechanism behind the development of junctional spinal disorders, especially at the lower end of constructs, still remains poorly understood.The presence of osteoporosis, multiple medical comorbidities, smoking, neurological disorders responsible for the deformity (such as Parkinson's disease), an aging and rigid spine with ligamentous degeneration, hip osteoarthritis, muscle fatty infiltration, and a severe stiffness mismatch between the bone and spinal implants all contribute to the challenge posed by the surgical treatment of patients with ASD [1].Mechanical complications related to long constructs represent the most frequently reported complications in ASD surgery (up to 30%) [2].
Junctional disorders above the construct, also known as proximal junctional kyphosis (PJK) or proximal junctional failure (PJF), represent a much dreaded, well-described, and frequent complication, while the majority of cases remain asymptomatic.Degeneration can also arise distally to the fusion in the form of distal junctional kyphosis (DJK) or failure (DJF) [3].While PJK is usually a more progressive and sub-clinical entity which occurs in 6.8% to 30% of cases, DJF is far less frequent, much more acute and spectacular in its presentation.Distal junction failure can also appear in the form of different clinical entities according to Arlet and Aebi [1], which can be any combination of progressive loss of lumbar lordosis, loss of disc height with disc degeneration, spinal stenosis, segmental instability, acute wedging in the distal junctional disc, fracture of the lower instrumented vertebra, and failure of the instrumentation at the most distal level (most frequently at L5).
Several strategies to prevent distal junctional failure in ASD surgery were proposed, such as avoiding ending a long construct at L5 or at S1 without protecting it with pelvic fixation [4], avoiding overcorrection (>50%), and incorporating the first lordotic disc into the fusion construct.Yet, there is a lack of evidence in the literature to support these.Our goal was to describe a case series of patients with a multilevel posterior lumbar spinal fusion requiring revision surgery for DJF, taking into account sagittal spinopelvic alignment, to identify potential clinical and radiographic risk factors, and finally propose prevention and revision strategies in light of the existing literature.

Study Design and Population
From June 2009 to January 2019, the data from 19 patients treated for DJF/DJK were retrospectively collected and analyzed in a single center.A local ethics committee approved this study.Inclusion criteria were (1) degenerative spinal deformity and/or (2) multilevel posterior spinal fusion, (3) requiring operative revision, and (4) six-month minimum follow-up.Demographic data were collected, such as age, gender, BMI, symptoms including abnormal neurological findings previous to treatment, number of revision cases, and relevant comorbidities, such as osteoporosis, smoking status, Parkinson's disease, alcohol abuse.Surgical data were also collected, including the surgical approach, type of instrumentation, presence of osteotomy and interbody fusion at the most caudal level, perioperative complications, and the need for revision surgery.All charts from all cases and complications were carefully analyzed and discussed with the operator in order to identify a potential cause for the failure.Eventually, a consensus was reached for each case by three attending surgeons.A radiographic analysis was conducted on preoperative standing X-rays (index procedure), postoperative standing X-ray (within 1 week of the index procedure), preoperative standing X-ray (before each revision), postoperative standing X-ray (within 1 week of each revision), and at final follow-up.The measured radiographic parameters were as follows: spinopelvic parameters (pelvic incidence, PI, lumbar lordosis, LL, sacral slope, SS, pelvic tilt, PT, PI-LL, sagittal vertical axis, SVA), lower instrumented vertebra (LIV), upper instrumented vertebra (UIV), number of fused levels, and the presence of an interbody spacer.

Definition of Distal Junctional Failure/Distal Junctional Kyphosis
DJK was defined as an abnormal distal junctional angle superior or equal to 10 degrees and at least 10 • greater than the preoperative value [3].The presence of both criteria was necessary to classify each event.The distal junctional angle was defined as the Cobb angle between the superior endplate of the LIV and the inferior endplate of the adjacent vertebra distal to the LIV (LIV + 1).This typically translated into loss of lumbar lordosis or disc degeneration, with loss of height.
Different types of DJF were identified: acute disc wedging distal to the instrumentation, fracture of LIV, osteoporotic fracture distal to the construct, failure of the instrumentation at the LIV, spinal stenosis, segmental instability distal to the construct [1].In this context, instability was defined as a difference of either 5 degrees or more in motion within a fusion segment observed between flexion and extension, or a vertebral body translation of 3 mm or more [5,6].

Potential Risk Factors
Potential risk factors screened in this case series [7,8]:

Statistical Analysis
Statistical analysis was performed using IBM SPSS Statistics software version 21.0 (IBM, Armonk, NY, USA).All data were expressed in mean values ± standard deviations.Comparisons between pre-and postoperative changes were evaluated using the Mann-Whitney U test.Statistical significance was assumed at p values of less than 0.05.

Results
The preoperative and postoperative clinical and radiographic data are summarized in Tables 1 and 2.

Population and Operative Data (Table 1)
Nineteen revision cases due to distal junctional failure after multilevel posterior spinal fusion surgery were reported.The mean age was 64.74 ± 13.55 years and the average followup was 4.7 ± 2.4 years.The sex ratio was 1.7:1, 12 women and seven men.There were eight cases of scoliosis or kyphoscoliosis (42.1%), two cases of trauma complicating elective spine surgery (10.5%), and five cases involving one or more degenerative spondylolisthesis (26.3%).The average number of instrumented levels was 6.79 ± 2.97.We observed that 84.2% (n = 16) of the patient sample had at least one known risk factor for mechanical complications.The most frequently observed LIV was L5 in 10 cases (52.6%), while two patients (10.5%) had a LIV at S1 without pelvic fixation.Only six patients (31.5%) had a circumferential fusion at most caudal level (TLIF only), performed during the index surgery.Reported initial DJFs were vertebral fracture distal to the fusion (n = 5), screw pull-out (n = 9), spinal stenosis (4), instability (n = 4), and one early DJK.Mechanical complications after a first revision for DJF included screw pull-out (4), screw fracture (3), non-union (2), and an iatrogenic spondylolisthesis at the distal end of the construct.Mechanical complications after a second revision for DJF included screw pull-out (2) and non-union (2).The average time to revision for DJF was 15.62 ± 17.86 months.Eight patients (42.1%) required a revision during their first postoperative year.The most frequently performed revision surgery was a distal extension of the fusion with iliac bolts.In this series, nine patients (47.3%)only required one revision.At the second revision stage, there were nine patients (47.3%).Finally, three required a third revision (15.7%).Iliac bolts were used only twice at the index procedure (10.5%), in conjunction with circumferential fusion at the distal end (L5-S1 TLIF).However, these two cases failed with failure at the iliac screws.Iliac bolts alone did not prevent failures.Indeed, they were either fractured (five times) or loosened/pulled out (six times).The use of iliac bolts was more prevalent in the revision cases: 12 cases (12/18, 66.6%) with iliac bolts at the first revision, 7 cases at the second revision (7/9, 77.7%), and 2 at the third revision (2/3, 66.6%).Success or failure can be defined by the need for subsequent revisions.In this case, regarding the outcomes of circumferential fusion (TLIF, ALIF, or LLIF) associated with pelvic fixation, this association was employed 12 times and resulted in mixed results (six failures and six successes).In order to prevent further revisions, the association of ALIF at L5-S1 and pelvic fixation was employed five times and provided the best results: four successes (no revisions to date) and one failure.

Radiographic Analysis (Table 2)
Despite the fact that before each surgery, most patients were sagittally malaligned (Figures 1A and 2B), revision surgery partially restored the SVA (Figures 1B and 2C), though this was not statistically significant (p > 0.05).During follow-up, the preoperative SVA increased with each complication and revision surgery.SVA correction decreased as well with each complication and revision surgery.Interestingly, each surgery tended to reduce the PI-LL mismatch, while this was statistically significant only for the index cases.Moreover, the patients revised twice had a significantly higher PI than the patients revised only once (p < 0.05) (Table 2).Lumbar lordosis correction increased with the number of revisions, though it did not reach statistical significance.Indeed, the average lumbar lordosis correction for patients with two failures or more was about three times higher than the average lumbar lordosis correction observed in the index case (p < 0.05) (Table 2).
Figures 1-3 illustrates some typical cases, and the management of DJF.reduce the PI-LL mismatch, while this was statistically significant only for the index cases.Moreover, the patients revised twice had a significantly higher PI than the patients revised only once (p < 0.05) (Table 2).Lumbar lordosis correction increased with the number of revisions, though it did not reach statistical significance.Indeed, the average lumbar lordosis correction for patients with two failures or more was about three times higher than the average lumbar lordosis correction observed in the index case (p < 0.05) (Table 2).

Discussion
Industrial societies have been challenged by an aging population with constantly higher functional demands [9].As a result, every year sees a further increase in adult spinal deformity cases [2].However, these cases are technically difficult due to advanced degenerative processes, osteoporosis, and weakened posterior elements (fatty degeneration of muscles).Pre-existing sagittal malalignment and co-existing hip and knee osteoarthritis may also add a level of complexity to these cases [10].All these factors contribute to a higher risk of postoperative malalignment, which is a well-described predictor of hardware failure [11][12][13].Because the vast majority of failures occur at the proximal end of spinal constructs, distal junctional failure has been relatively poorly described to this date.Obviously, a common factor in these cases is a high mechanical stress at the lower end of the construct.Our belief is that in most cases, a local or regional angular abnormality leads to postoperative malalignment, which in turn becomes the genesis of high mechanical stress at both upper and lower extremities.In order to reduce that stress, the most ergonomic and energy-efficient strategy is unconsciously or semi-consciously chosen by patients above and below the fusion (pelvic retroversion if hips joints are not osteoarthritic, thoracic hypokyphosis, knee flexion) [14].The lower energy option is often the first chosen.When these mechanisms are overrun, the energy can be restituted in various forms (screw pull-out, fracture, dislocation, rod or screw fracture) [15].
These results suggest that DJF is usually not an isolated event.It behaves similar to a mechanical cascade that can only be stopped surgically.Indeed, the majority of our

Discussion
Industrial societies have been challenged by an aging population with constantly higher functional demands [9].As a result, every year sees a further increase in adult spinal deformity cases [2].However, these cases are technically difficult due to advanced degenerative processes, osteoporosis, and weakened posterior elements (fatty degeneration of muscles).Pre-existing sagittal malalignment and co-existing hip and knee osteoarthritis may also add a level of complexity to these cases [10].All these factors contribute to a higher risk of postoperative malalignment, which is a well-described predictor of hardware failure [11][12][13].Because the vast majority of failures occur at the proximal end of spinal constructs, distal junctional failure has been relatively poorly described to this date.Obviously, a common factor in these cases is a high mechanical stress at the lower end of the construct.Our belief is that in most cases, a local or regional angular abnormality leads to postoperative malalignment, which in turn becomes the genesis of high mechanical stress at both upper and lower extremities.In order to reduce that stress, the most ergonomic and energy-efficient strategy is unconsciously or semi-consciously chosen by patients above and below the fusion (pelvic retroversion if hips joints are not osteoarthritic, thoracic hypokyphosis, knee flexion) [14].The lower energy option is often the first chosen.When these mechanisms are overrun, the energy can be restituted in various forms (screw pull-out, fracture, dislocation, rod or screw fracture) [15].
These results suggest that DJF is usually not an isolated event.It behaves similar to a mechanical cascade that can only be stopped surgically.Indeed, the majority of our patients had more than one revision.The SVA correction tended to follow a law of diminishing returns after each mechanical complication and therefore revision.In addition, the lumbar lordosis average correction increased with each round of revision, which shows that more effective corrective osteotomies were employed after each failure.The fact that the PI was higher in patients with two revisions or more suggests that high sheer forces at the distal end of the construct increases the risk of DJF and renders revisions more challenging.
Regarding the treatment pattern, there was a general tendency to reinforce the distal end of the construct with either an interbody device for circumferential fusion or iliac bolts, or both (Table 1).If there was a TLIF at the most caudal level and failure occurred, an ALIF was performed.If the ALIF failed, it was revised with iliac bolts.Only six patients had a distal circumferential fusion at the index procedure.Retrospectively, it appears as probably too few, especially at the distal end.Interestingly, there was no ALIF as a means of circumferential fusion in the index procedure, which is the result of surgeons' habits.Only TLIFs were used for that purpose.ALIFs were reserved for revisions.Interestingly, the combination of an ALIF at L5-S1 with pelvic fixation appeared to arrest the progression of the "revision cascade".This may be explained by the lordotic effect of the L5-S1 ALIF cages, which in theory can more easily correct SVA since they are located at the bottom of the construct.Because of the immense constraints in long fusions and obese patients particularly, it seems retrospectively logical to protect iliac bolts with a lumbosacral fusion anteriorly.Yet, this may have been falsely considered a success due to a lack of follow-up.
Failure is not exclusively the result of instrumentation.It can also result from soft tissue damage.Indeed, the posterior tension-band was damaged in three cases in this series.It is probable that this played a role in the development of DJF.Furthermore, removing instrumentation at the distal end on a previous fusion probably resulted in failure in case 1.
In most cases, a lesson can be learnt from a failure, and this is the reason why the authors chose to share this case series with esteemed peers.Failures do not occur randomly.There are patterns of techniques and patients that lead to them.In addition, the lever arm represented by long instrumentations is usually greatest on the lower end of the constructs, which may explain the «spectacular» and early hardware failures defined as distal junctional failures.Poor understanding of the etiology of a hardware failure will ultimately lead to another disaster.Screws do not always pull-out because of poor bone density, and augmented screws with PMMA are not the sole answer for the prevention of junctional failures.Obviously, if a patient is significantly malaligned immediately postoperatively, any surgical intervention aimed at reducing the rate of junctional failure is bound to fail (screws may pull out along with their cement block or fracture the entire vertebra; fractures or dislocations occur in vertebrae peripheral to the instrumentation despite vertebroplasty at junctional levels).This vicious circle can be broken once the extremities are off-loaded, i.e., when the patient is adequately aligned.Several risk factors for DJF were described in the literature [7,8,15].
They can be separated into two main categories: factors that can be influenced by the surgeon (sagittal imbalance, overcorrection, lower instrumented vertebra selection).DJF may result in imbalance and unacceptable deformity.It is not always detectable clinically, and it is often not predictable from the analysis of preoperative radiographs [16,17].Furthermore, overcorrection may occur in younger patients (children with adolescent idiopathic scoliosis or Scheuermann's disease).Preoperative thoracolumbar kyphosis is a risk factor for the development of postoperative DJK.Similarly, the presence of increased kyphosis after surgery in the T10-L2 region constitutes a "risk factor" for the development of DJK [16].The LIV selection is also critical, especially in Scheuermann's disease.In this scenario, stopping short of the FLD or not including the stable sagittal vertebrae (SSV) or a high residual LIV plumb line will lead to a higher risk of DJK and DJF [1].In the case of DJK, these patients should receive more attention and be scheduled for continuous follow-up in the first postoperative years [18].Another group of DJF risks factors comprise those that cannot be influenced by the surgeon (high BMI, osteopenia/osteoporosis, older age, substance abuse, multiple previous surgeries, and concomitant hip osteoarthritis).Whenever they are modifiable, adequate treatment should be initiated in order to mitigate the risk of postoperative DJF.
Multiple prevention tools are available to surgeons.Matching lumbar lordosis to pelvic incidence and respecting the global sagittal alignment is one of them [19,20].Another issue is that lordosis mostly affects the L4-S1 segment rather than the L1-L4 segment.Finally, respecting the posterior tension band, as it was shown in PJK or PJF prevention [8], including the first lordotic disc in a thoracic kyphosis correction [1], choosing the right instrumentation stiffness adapted to the anticipated post-correction spine stiffness (rod diameter, metal type) [21], creating a "soft-landing" with sublaminar bands [22] or hooks at the extremities of fusion constructs (as can be used for PJK and PJF prevention) [23] are other commonly described tips and tricks, though there is no consensus or even high evidence backing their use.Because most of these factors were described in the setting of the research on PJK and PJF, using these principles in a similar manner at the distal end of constructs may not necessarily prove to be adequate in the future.Further specific research is needed for DJF.
In addition, it is wise to analyze the lumbosacral "base" of a construct following a PJF event.Indeed, our belief is that the same reasons (sagittal malalignment, often between L4 and S1) can alternatively lead to PJF or DJF.This has yet to be demonstrated.
Conflicting evidence exists for the choice of the LIV to prevent DJK.Yang et al. [24] showed that stopping at or below the SSV nearly eliminates the risk of DJK in the case of AIS with selective fusions.Yet, this was achieved at the expense of incorporating additional fusion levels.What typically happens at L5-S1 post-fusion is another controversial topic.Cho et al. reported that subsequent advanced L5-S1 disc degeneration occurred in 58% of the patients whose fusion ended at L5 [25].Again, these studies focus on DJK and typically do not address DJF.Fusing on S1 would eliminate the possibility of L5-S1 disc degeneration and loss of lordosis, but it is also associated with a higher rate of pseudarthrosis.Yasuda et al. recommended spinopelvic fixation using iliac bolts for long corrective fusions in the setting of adult spinal deformity surgery, given the high failure rate in patients with long fusions when stopping at L5 or S1 in their series [4,25].The current work emphasizes the importance of lower lumbar lordosis correction (L4-S1) with anterior support (tall interbody devices with significant built-in lordosis) to increase the fusion rates and achieve satisfactory postoperative balance.Indeed, the best chances of reaching optimal sagittal alignment seems to lie below the instrumentation and not above it.In the case of multiple revisions, it is obvious that addressing only the symptom (for example toggling screw treated with percutaneous vertebroplasty in the pelvis or vertebra) will not arrest the underlying mechanical issue.Avoiding the repetition of similar mechanical mistakes, preoperative surgical planning (stiffness, alignment, screw type, cementation), preparing the patient to control non-surgeon-dependent DJF risk factors (BMI, T-score, smoking cessation, osteoporosis treatment, evaluation and treatment of hip osteoporosis) [26] are not the unique keys to success but help avoid taking the highway to failure.More recently, McDonnell et al. demonstrated in a systematic review on DJF following ASD surgery that stopping at L5 was risk factor for developing DJF postoperatively and that there could be a protective effect of anterior-posterior fusions on postoperative DJF [27].In a subsequent study, they showed that performing a pedicle subtraction osteotomy, undercorrecting lumbar lordosis and sagittal vertical axis, are significantly associated with postoperative DJF [28].The role of sagittal balance was again shown to be of importance by Montanari et al. in their retrospective case-control cohort study of posterior spinal fusions.Indeed, they found an association between DJF and age over 40 years, as well as a lack of restoration of PT, PI-LL, and T1 pelvic angle (TPA) [29].
A simple algorithm for DJF management is proposed in Figure 4.

Limitations of This Study
The number of patients is low and reflects failures of multilevel posterior lumbar fusion surgery in a single center over a 10-year period.Given the fact that DJF is a rare event, this study is most likely too under-powered to detect new risk factors.During the study period, there were approximately 7500 spine fusions performed in our department.The focus was on DJF cases that required a revision surgery.The global incidence was therefore 0.25% for all fusions performed, which includes non-ASD surgeries.A multi-center and controlled prospective study would have overcome this obstacle.However, surgeons are usually reluctant to report their worst complications.It was therefore decided to proceed with a single-center study only and raise awareness about the cascade of complications generated by DJF.The retrospective nature of this study and the absence of controls also has inherent limitations.Yet, there is great interest in defining the patterns between two different types of DJF: those that are "cured" and those that recur.

Limitations of This Study
The number of patients is low and reflects failures of multilevel posterior lumbar fusion surgery in a single center over a 10-year period.Given the fact that DJF is a rare event, this study is most likely too under-powered to detect new risk factors.During the study period, there were approximately 7500 spine fusions performed in our department.The focus was on DJF cases that required a revision surgery.The global incidence was therefore 0.25% for all fusions performed, which includes non-ASD surgeries.A multicenter and controlled prospective study would have overcome this obstacle.However, surgeons are usually reluctant to report their worst complications.It was therefore decided to proceed with a single-center study only and raise awareness about the cascade of complications generated by DJF.The retrospective nature of this study and the absence of controls also has inherent limitations.Yet, there is great interest in defining the patterns between two different types of DJF: those that are "cured" and those that recur.

Conclusions
The current data analysis suggests that female gender, osteoporosis, stopping at L5 or at S1 without iliac fixation in long constructs with UIV proximal to L1 represent high risks of DJF.Insufficient sagittal alignment restoration also probably represents a high risk of DJF and revision surgery.The association of an anterior lumbosacral fusion with pelvic fixation appeared to decrease the risk of the recurrence of DJF.
The main goals of the treatment of DJF are to restore spinal balance, through a posterior approach (hardware revision, correction, treating non-union) and to obtain a solid anterior fusion with a wide and lordotic spacer (either through a traditional anterior retroperitoneal approach for an anterior lumbar interbody fusion, ALIF, or a minimally invasive lateral approach such as a lateral lumbar interbody fusion, LLIF).
This series shows that this type of circumferential fusions may help maintain surgical

Conclusions
The current data analysis suggests that female gender, osteoporosis, stopping at L5 or at S1 without iliac fixation in long constructs with UIV proximal to L1 represent high risks of DJF.Insufficient sagittal alignment restoration also probably represents a high risk of DJF and revision surgery.The association of an anterior lumbosacral fusion with pelvic fixation appeared to decrease the risk of the recurrence of DJF.
The main goals of the treatment of DJF are to restore spinal balance, through a posterior approach (hardware revision, correction, treating non-union) and to obtain a solid anterior fusion with a wide and lordotic spacer (either through a traditional anterior retroperitoneal approach for an anterior lumbar interbody fusion, ALIF, or a minimally invasive lateral approach such as a lateral lumbar interbody fusion, LLIF).
This series shows that this type of circumferential fusions may help maintain surgical correction, increase the rate of fusion, and reduce the overall longer-term mechanical complications.The efficacy of PSO should be balanced with its high risk of complications reported in the literature.It is to be avoided as a treatment for sagittal imbalance below a long fusion with DJF.Yet, it may still be an option in the case of an iatrogenic flatback and DJF.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Figures 1 -
Figures 1-3 illustrates some typical cases, and the management of DJF.

Figure 4 .
Figure 4. Proposed algorithm for the management of DJF based on clinical symptoms and associated diagnoses (spinal stenosis, sagittal imbalance, non-union).

Figure 4 .
Figure 4. Proposed algorithm for the management of DJF based on clinical symptoms and associated diagnoses (spinal stenosis, sagittal imbalance, non-union).

Table 1 .
Main demographic and surgical data from 19 cases of distal junctional failure.