Etiology of Chronic Low Back Pain in Patients Having Undergone Lumbar Fusion

Abstract

Objective. To estimate the prevalence of lumbar internal disc disruption, zygapohyseal joint pain, sacroiliac joint pain, and soft tissue irritation by fusion hardware in post-fusion low back pain patients compared with non-fused patients utilizing diagnostic spinal procedures.

Design. Retrospective chart review.

Setting. University spine center.

Patient Sample. Patients presenting to a community-based, multidisciplinary, academic spine center (65.9% female, mean age 54.4 years, median pain duration 12 months).

Interventions. Charts of consecutive low back pain cases completing diagnostic spinal procedures including provocation discography and zygapohyseal joint, sacroiliac joint, and fusion hardware blockade were retrospectively reviewed.

Outcome Measures. Based on the results of discography and/or diagnostic blockades, subjects were classified with internal disc disruption, zygapohyseal joint pain, sacroiliac joint pain, or fusion hardware related pain.

Results. The diagnoses of 28 fusion cases identified from 170 low back pain patients undergoing diagnostic procedures included 12 with sacroiliac joint pain, seven with internal disc disruption, five with zygapohyseal joint pain, and four due to soft tissue irritation from fusion hardware. No significant differences were noted in zygapohyseal joint mediated pain with and without fusion history. Mean ages of patients were similar with and without fusion history for cases diagnosed as internal disc disruption.

Conclusion. In patients' recalcitrant to non-interventional care, the sacroiliac joint is the most likely source of low back pain after lumbar fusion followed by internal disc disruption, zygapohyseal joint pain, and soft tissue irritation due to fusion hardware. Sacroiliac joint pain is more common after fusion, while internal disc disruption is more common in non-fusion patients.

Lumbar, Fusion, Intervertebral Disc, Facet Joint, Sacroiliac Joint, Chronic Pain

Introduction

As recently as 5 years ago, approximately over 700,000 lumbar spine fusion surgeries were performed in the United States annually [1]. A respectable proportion of these patients will experience persistent or new low back pain (LBP) post-operatively due to various etiologies [2–9]. Collectively, cases of recurrent LBP and/or lower limb pain after lumbar fusion have been referred to as “Failed Back Surgery Syndrome”[3–9]. However, such a phrase is non-descriptive and not diagnostic [8]. Potential explanations for LBP after lumbar fusion include pseudoarthrosis, discogenic pain at a previously fused [10] or adjacent level, adjacent level zygapohyseal joint arthropathy (ZJA), sacroiliac joint dysfunction (SIJD) [8,11], or fusion hardware related LBP [12]. Other conditions such as epidural/perineural fibrosus, arachnoiditis, or latent spinal stenosis [8,11] would present with more significant lower limb symptoms than axial LBP [8].

Frymoyer et al. concluded three decades ago that sacral sulcus pain encountered in 37% of LBP patients after lumbar fusion was iliac graft donor site related, after discarding the sacroiliac joint (SIJ) as the etiologic source due to lack of objective physical examination and imaging findings [10,11]. Ebraheim et al. later insinuated that direct SIJ injury can occur during iliac crest graft harvest resulting in symptomatic ligamentous disruption and degeneration after fusion [12]. More recently, Katz et al. retrospectively identified definite SIJ pain in 32% of LBP patients who had a history of both lumbosacral fusion and completion of post-operative SIJ diagnostic blocks [13]. No correlation was reported to exist between the side of LBP and side of graft harvest rendering direct SIJ damage after graft harvesting improbable.

By employing diagnostic SIJ blocks, Maigne and Planchon demonstrated a 35% prevalence of SIJ pain after fusion [14]. The investigators found that posterior iliac bone graft harvesting was not significantly related to the presence of SIJ pain; but, that a history of post-operative LBP pattern different from the preoperative LBP pattern was significantly related to SIJ pain. Furthermore, a time period greater than 3 months between fusion and onset of the different LBP was indicative (P = 0.17) but not predictive. Maigne et al. suggested altered biomechanical loads subsequent to the fusion might stimulate onset of symptoms. Although not significant, more SIJ cases were observed by Maigne and Planchon in patients fused to the sacrum than not fused to the sacrum. Maigne's work corroborates others' suggestion that iliac crest donor site mediated LBP and SIJ mediated LBP are separate, independent conditions.

Alternate spinal structures have been recognized as sources of LBP after fusion. Weatherly et al. [15] first documented evidence of painful discs within a solid fusion. Barrick et al. subsequently confirmed this scenario in a larger case series [16]. These investigators noted discogenic LBP within a solid posterolateral fusion utilizing provocation discography, ruling out discogenic pain from adjacent and remote levels. Reduction in LBP and patient disability occurred in this patient cohort after incorporation of a technically acceptable anterior interbody fusion at the index level. Fusion hardware mediated LBP has been described [17]. Presumably, either mechanical or material irritation of the soft tissues adjacent to the hardware results in symptoms [17]. Removal of the hardware can reduce LBP in affected individuals [17]. Zygapohyseal joints (ZJs) have not been investigated as a source of chronic LBP after fusion but are recognized as potential pain generators.

In patients with chronic LBP without a history of lumbar fusion, the prevalence of painful internal disc disruption (IDD) has been reported as 39–43% [13,14], painful ZJA as 15–31% [14–16], and painful SIJD as 13–18% [14,17,18]. The prevalence of symptomatic SIJ has been reported to increase to 32–37% in lumbosacral fusion patients [19–21]. The prevalence of IDD, symptomatic ZJA, and fusion hardware LBP as well as corroboration of painful SIJD in lumbar fusion patients has not been published. The purpose of this study is to estimate the prevalence of lumbar IDD, symptomatic ZJA, painful SIJD, and soft tissue irritation by fusion hardware in consecutive LBP patients with a history of fusion having completed precise, controlled diagnostic spinal procedures and to compare these prevalence estimates to those in non-fusion chronic LBP patients.

Methods

After obtaining Institutional Review Board approval, charts from consecutive LBP patients were reviewed. These were LBP patients presenting to a community-based, multidisciplinary, academic spine center. Patients were referred to the spine center from community and university spine surgeons (neurosurgery and orthopedics), physiatrists, non-spine surgeons, primary care physicians, rheumatologists, endocrinologists, neurologists, and occupational health physicians. Enrolled cases comprised of LBP patients suffering from chronic LBP recalcitrant to conservative care (spine focused physical therapy, oral analgesics, and oral anti-inflammatory medications) who underwent precision, diagnostic spinal procedures.

Each patient either underwent provocation lumbar discography (PLD), dual diagnostic ZJ blocks (ZJBs) with local comparative anesthetics, intra-articular diagnostic sacroiliac joint injections (SIJBs), or injection of anesthetic onto putatively painful posterior fusion hardware, depending on the clinical presentation. Some subjects underwent multiple diagnostic procedures until the source of their LBP was identified. If the initial diagnostic procedure was negative, the next most likely structure in the diagnostic algorithm was interrogated. For ethical reasons, once a source of the subject's LBP was identified, subsequent diagnostic procedures were not performed in order to implement definitive treatment.

Patients reporting paravertebral LBP without midline LBP [22,23], which was exacerbated by standing and/or walking [24], and who demonstrated ≤2 positive SIJ provocative maneuvers [25] and/or lack centralization during the McKenzie evaluation [26] typically underwent ZJB first, followed by SIJB, and then PLD if the preceding diagnostic procedure was negative. The side and joint level selected by pain referral pattern [27,28] were investigated first moving from most likely to less likely ZJ level. Patients reporting paravertebral LBP without midline LBP [22,29–34] and 3 positive out of 5 SIJ provocative maneuvers [25,35] without centralization during the McKenzie evaluation [26] underwent SIJB, followed by ZJBs, and then PLD unless the initial diagnostic blocks were positive. Patients reporting paravertebral LBP with a previous history of posterior fusion with pedicle screws and hardware whose LBP was reproducible by single-digit palpation over the hardware underwent diagnostic blockade of the hardware in a triple blockade fashion using 2% lidocaine first, then 0.5% marcaine second, followed by a placebo injection. Patients reporting midline LBP with or without paravertebral LBP, centralization during the McKenzie evaluation [26], and/or LBP during sustained hip flexion [36] underwent PLD, initially followed by ZJB or SIJB if discography was negative.

Positive discography was defined as concordant/partial concordant LBP (>6/10) at low pressure (<50 psi over opening pressure) due to ≥grade III annular tears [37–39]. Diagnostic blockade of ZJ, SIJ, or other structures was deemed positive if the patient's index pain was relieved by ≥75% after injection of each anesthetic [22,29–34]. In the case of fusion hardware blockade, minimal relief after the placebo injection was required to constitute a positive block. Initial ZJBs were performed by intra-articular injection of anesthetic followed by confirmatory medial branch blockade.

Based on the results of PLD and/or diagnostic blockades, subjects were classified as having IDD, ZJA, SIJ, or fusion hardware related pain. Charts of patients who did not undergo definitive diagnostic procedures due to clinical improvement of LBP were reviewed but not enrolled in this study. SAS version 9.2 (SAS Institute Inc., Cary, NC) was used for all data analysis. Descriptive statistics such as means and standard deviations (SDs) for continuous data and counts and percentages for nominal data were computed. Chi-square tests, t-tests, and Wilcoxon rank-sum tests were used to compare proportions and means/medians between groups using a significance level of α = 0.05.

Results

A total of 378 cases from 358 patients seen between November 2007 and December 2008 were reviewed. Cases reviewed were primarily female (65.6%), with a mean age of 52.3 years (SD = 15.3) and median duration of LBP of 12 months (interquartile range [IQR] = 6–24). There were 208 cases from 198 patients not included in subsequent calculations because these patients did not undergo definitive diagnostic procedures. Of the 170 cases from 160 patients having definitive diagnostic procedures, 65.9% were female, the mean age was 54.4 years (SD = 16.2) and the median duration of LBP was 12 months (IQR = 6–32).

There were 28 cases (16.5%) undergoing definitive diagnostic procedures that had a history of fusion, 141 cases with no history of fusion, and for the remaining one case, the history of fusion was unknown (see Table 1). There was no evidence of statistically significant differences in gender (P = 0.0557), age (P = 0.4990), or duration of LBP (P = 0.7194) between cases with and without fusion history. Four of the fusion cases (14%) had undergone fusion incorporating an anterior construct, 23 (82%) had undergone posterolateral fusion without an anterior construct, and one (4%) was status post Harrington rod placement for deformity correction. Eleven cases (39%) were 1 level, seven (25%) were 2 level, six (21%) were 3 level, and four (14%) had undergone fusion of 4 or more levels. The distribution of the specific levels fused is summarized overall and by source of LBP in Table 2.

Table 1

Characteristics of cases with and without history of fusion

	History of Fusion
	Yes (N = 28)	No (N = 141)
Gender, N (% female)	14 (50.0)	97 (69.0)
Age (years), mean (SD)	56.3 (13.7)	54.0 (16.7)
Duration of LBP (months), median (IQR)	12 (5.5–60)	12 (6–24.5)

SD = standard deviation; LBP = low back pain; IQR = interquartile range.

Table 2

Distribution of the levels fused by source of low back pain

Levels Fused	Total		IDD		ZJA		SIJ		Hardware
	N	Percent	N	Percent	N	Percent	N	Percent	N	Percent
L2-5	2	7.1	1	14.3	—		—	—	1	25.0
L3-5	2	7.1	—	—	1	20.0	1	8.3	—	—
L4-5	5	17.9	1	14.3	2	40.0	1	8.3	1	25.0
L2-S1	2	7.1	1	14.3	—	—	1	8.3	—	—
L3-S1	4	14.3	1	14.3	—	—	3	25.0	—	—
L4-S1	5	17.9	1	14.3	—		4	33.3	—	—
L5-S1	6	21.4	2	28.6	1	20.0	2	16.7	1	25.0
T-L4	2	7.1	—	—	1	20.0	—	—	1	25.0

IDD = internal disc disruption; ZJA = zygapohyseal joint arthropathy; SIJ = sacroiliac joint.

SIJs were symptomatic in 12 of the fusion cases (43%), seven cases (25%) had IDD, five cases (18%) had painful ZJA, and four cases (14%) of soft tissue irritation by fusion hardware were detected. Table 3 summarizes these prevalence rates and 95% confidence intervals (CIs) along with the mean ages for the cases by source of LBP. Ten of the 12 SIJ cases with a fusion history had fusion to the sacrum (83%; 95% CI = 55%, 95%), and the remaining two cases had fusion to L5 (see Table 2). Two of the seven IDD cases had a painful disc at a previously posteriorly fused level and four of the seven cases directly adjacent to a fused level. One of the seven IDD cases had a painful disc 1 level removed from previously fused levels (L2-L3 s/p L4-S1 fusion) (see Table 4). Four of the five ZJA cases had a symptomatic ZJ immediately adjacent to a previously fused level (see Table 5).

Table 3

Prevelance and mean age by source of low back pain after fusion

		Prevalence		Age
Source of LBP	N	Percent	95% CI	Mean	SD	95% CI
IDD	7	25.0	(12.7, 43.4)	46.1	6.5	(40.1, 52.2)
ZJA	5	17.9	(7.9, 35.6)	57.6	13.1	(41.3, 73.9)
SIJ	12	42.9	(26.5, 60.9)	61.0	12.2	(53.2, 68.8)
Hardware	4	14.3	(5.7, 31.5)	58.5	22.3	(23.0, 94.0)

CI = confidence interval; SD = standard deviation; LBP = low back pain; IDD = internal disc disruption; ZJA = zygapohyseal joint arthropathy; SIJ = sacroiliac joint.

Table 4

Symptomatic IDD levels in seven cases after fusion

Levels Fused	Painful Level
L2-5	L5-S1
L4-5	L4-5
L2-S1	L1-2
L3-S1	L2-3
L4-S1	L2-3
L5-S1	L4-5
L5-S1	L5-S1

IDD = internal disc disruption.

Table 5

Symptomatic ZJA levels in five cases after fusion

Levels Fused	Painful Level
L3-5	L5-S1
L4-5	L5-S1
L4-5	L5-S1
L5-S1	L4-5
T-L4	L5-S1

ZJA = zygapohyseal joint arthropathy.

IDD occurred as a source of LBP with significantly greater frequency in patients without fusion history than with a history (45.4% vs 25.0%; P = 0.0406). Cases of SIJ pain were more common for subjects with a fusion history than without (42.9% vs 12.8%; P = 0.0005); there were not significant differences in ZJA sources of LBP between subjects with and without fusion history (33.3% vs 17.9%; P = 0.0911). Mean ages of LBP patients were similar between cases with and without fusion history for LBP cases diagnosed as IDD (46.1 vs 43.4 years), ZJA (57.6 vs 59.9 years), and SIJ pain (61.0 vs 62.7 years).

Discussion

Our findings suggest that the specific etiology of chronic LBP after a lumbar fusion can be identified. However, our results do not allow us to support or discard the notion that multiple painful structures can coexist after lumbar fusion. Each patient whose LBP was diagnosed as discogenic after positive provocation discography did not necessarily undergo negative joint blockade. Nonetheless, we can assert that patients complaining of LBP after lumbar fusion typically do suffer such symptoms due to involvement of a spinal structure within or adjacent to the index fusion segment. The most common source of which is the SIJ followed, respectively, by the intervertebral disc(s), ZJ(s), and soft tissue irritation by fusion hardware.

LBP symptoms emanating from intra-articular SIJ pathology should be recognized as a diagnostic group separate from iliac donor site mediated LBP. The prevalence of SIJ pain in our cohort was 43%, yet, may be as high as 61%. It is likely that SIJ pain after lumbar fusion represents a third to over half of chronic LBP cases having undergone previous fusion. Greater than 80% of our subjects having a history of both a fusion and a subsequent SIJ pain, had fusion to the sacrum (P = 0.0032). A risk factor for development of SIJ pain after fusion is inclusion of the sacrum into the surgical construct. Our finding corroborates earlier reports of the correlation between SIJ pain and fusion to the sacrum [21,40]. Computed sacrum angular motions and average stress on SIJ articular surfaces increase after lumbar fusion, especially fusion to the sacrum [41]. It is logical that increased motion and stress across the SIJ instigates intra-articular derangement leading to persistent LBP.

Discogenic LBP, ZJA, and fusion hardware mediated LBP occur less commonly than SIJ pain in fusion patients (Table 3) but should not be dismissed as clinically insignificant. Fusion by posterior fixation reduces intradiscal pressure up to 55% within the disc(s) of the fused segment [42] while increasing intradiscal pressure by 45% in adjacent level discs [42–44]. Expansion of the fusion across two segments increases adjacent level intradiscal pressure more so compared with adjacent to single level [43,44]. Incorporation of an anterior interbody fusion construct similarly increases adjacent level intradiscal pressure [44,45]. Regardless of fusion technique, loss of segmental motion due to instrumentation is compensated for by incremental increase in spine range of motion at adjacent levels [45]. Alterations in disc loading are associated with impaired intradiscal metabolism due to inefficient nutrient and by-product transport into and out of the disc [46–48] inducing a catabolic state resulting in disc degeneration [42,49]. Adjacent level zygapohyseal capsular strain during physiological motions is increased after anterior spinal fixation, which, if suprathreshold for capsule nociceptors, may trigger LBP [50]. Zygapohyseal contact force has been observed to increase at levels adjacent to instrumented fusion [45], although not statistically significant, these changes were measured compared with intact, non-fused spines [45]. Fusion hardware itself has been associated with focal inflammation, granuloma formation [20,21,40], and bursae formation adjacent to the implant [18,19].

Force and load transfer to non-fused structures presumably plays a causal role in tissue injury leading to symptom manifestation. Although painful SIJs surpass painful lumbar intervertebral discs as the spinal etiology most frequently responsible for chronic LBP after fusion, lumbar IDD appears to still occur within fused segments despite internal fixation presumably due to persistent micro-motion undetected on imaging [16]. Lumbar IDD at adjacent and remote levels after fusion occurs as a result of undue strain and increased intradiscal pressures experienced by these discs after fusion [42–45]. Similarly, ZJA develops subsequent to capsular strain and intra-articular tissue damage related to increase load and articular surface friction [50]. Our clinical findings support the published biomechanical studies that demonstrate potentially injurious conditions exist for adjacent level structures after lumbar fusion. Our preliminary findings suggest that SIJ pain swaps places with IDD as the leading source of chronic LBP after fusion relative to non-fused LBP patients. However, our data allow us to comment with caution how changes in internal biomechanics might predict which spinal structure will likely become symptomatic. Aside from the correlation between fusion to the sacrum and SIJ pain, our data preclude commentary regarding how fusion technique (anterior–posterior vs posterolateral) and number of levels fused relate to IDD vs ZJA as the source of chronic LBP status post-fusion. Presumably, altered spine biomechanics play a role in these relationships because the mean age of each diagnostic group (SIJ, IDD, ZJA) were not statistically different between LBP patients with and without a history of lumbar fusion.

Internal stabilization via fusion surgery categorically changes the likely source of chronic LBP compared with non-fused LBP patients. The prevalence of SIJP is only 13% in non-fused LBP patients but becomes the most common source of LBP after a lumbosacral fusion with a prevalence of 43%. The prevalence of lumbar IDD decreases from 45% in LBP patients to 25% in LBP patients with a fusion. Painful ZJA prevalence is not significantly altered in those LBP patients with a lumbar fusion. Age does not appear to affect the prevalence of each diagnosis suggesting that other variables (altered biomechanics, smoking, etc.) are more influential in why certain spinal structures become symptomatic. IDD, including discs both outside and within the fusion, remains a common source of chronic symptoms in LBP patients after fusion, especially in the absence of an anterior construct. LBP and/or sciatic pain associated with adjacent segment degeneration sufficient to warrant surgical intervention occurs in a third of fused patients at 10 years [51,52]. In a randomized controlled trial, Ekman et al. [52] found adjacent segment degeneration occurs with a 38% increased frequency after lumbar fusion, especially with laminectomy, when compared with their non-fused counterparts. Less significant postoperative clinical improvement (pain and disability) occurred after fusion in patients with adjacent level degeneration compared with those fused without adjacent level degeneration [52]. In the LBP cases, the clinical diagnosis was not entirely clear as the presurgical planning centered more so on radiographic parameters and the presence of symptoms [52]. Symptomatic structures such as the SIJ not targeted by the investigators' fusion may explain some of the poorer outcomes. Our data support Frymoyer et al.'s initial impression, if not their results, to include the SIJ as a putatively painful structure in the differential for LBP after lumbar fusion.

For ethical reasons, each patient reviewed in our study did not undergo all diagnostic procedures (discography and diagnostic blocks). Rather, a focused, pragmatic algorithmic approach was employed. One could argue that an erroneous calculation of the prevalence estimate for lumbar IDD, ZJA, and SIJP was committed. By not performing discography on every patient, it is plausible that we failed to detect all cases of IDD and have underreported it. A similar comment could be made about diagnostic ZJ and SIJ blocks. However, each patient analyzed underwent definitive diagnostic procedures until we reached confirmation of the source of that patient's LBP. If a patient was initially evaluated with diagnostic ZJ and/or SIJ blocks that were negative, that patient underwent discography to verify the presence of IDD and vice versa. Only patients whose clinical status improved with proper care did not undergo diagnostic procedures. These cases were not included in our post-fusion LBP analysis.

Opponents of discography and, to a lesser degree, diagnostic procedures in general, could contend that false positive rates have overestimated our prevalence estimates. Application of meticulous technique and strict adherence to supported operational criteria for discography [37] will minimize false positive rates to acceptably low levels [38] allowing accurate detection of symptomatic lumbar IDD [31,37,53,54]. Similarly, sufficiently performed diagnostic ZJ blocks and SIJ injections are associated with acceptable false positive rates. Last, if our findings were skewed by false positives, we would have likely observed different prevalence data less congruent with previous reports. By virtue of the fact that most previously reported prevalence estimates for each diagnostic group (IDD, ZJA, SIJP) fall within our CIs for each group, our findings are consistent with previously published data despite this being a single center study with a small sample size.

Conclusion

Despite these shortcomings, this retrospective audit is the first preliminary study of the etiologies of chronic LBP after fusion in consecutive patients. In addition, our study establishes the notion that the prevalence rates of the sources of chronic LBP are altered by the history of a lumbar fusion. Our findings provide reason to further pursue subsequent studies investigating correlations between type of fusion and number of levels fused and sources of post-fusion LBP. Furthermore, our work corroborates previous publications attesting to the role of the SIJ in chronic post-fusion LBP. Last, implications of these clinical investigations may help refine strategies to minimize disruption of spinal biomechanics subsequent to spinal fusion.