an institutional six-year trend analysis of the

Spine www.spinejournal.com E1483

HEALTH SERVICES RESEARCH

SPINE Volume 38 , Number 23 , pp E1483 - E1490 ©2013, Lippincott Williams & Wilkins

An Institutional Six-year Trend Analysis of the Neurological Outcome After Lateral Lumbar Interbody Fusion

A 6-Year Trend Analysis of a Single Institution

Alexander Aichmair , MD , * Marios G. Lykissas , MD, PhD , * Federico P. Girardi , MD , * Andrew A. Sama , MD , * Darren R. Lebl , MD , * Fadi Taher , MD , † Frank P. Cammisa , MD , * and Alexander P. Hughes , MD *

DOI: 10.1097/BRS.0b013e3182a3d1b4

Study Design. Retrospective case series. Objective. To evaluate the proportional trend over time of neurological defi cits after lateral lumbar interbody fusion (LLIF) at a single institution. Summary of Background Data. Because lumbar nerve roots converge to run as the lumbar plexus within or less frequently underneath the posterior part of the psoas muscle, they are prone to iatrogenic damage during the transpsoas approach in LLIF, and adverse postoperative neurological sequelae remain a major concern. Methods. The electronic medical records and offi ce notes of 451 patients who had consecutively undergone LLIF between March 2006 and April 2012 at a single institution were retrospectively reviewed for reports on postoperative neurological defi cits. Results. A total of 293 patients (173 females and 120 males) met the study inclusion criteria and were followed postoperatively for a mean period of 15.4 ± 9.2 months (range: 6–53 mo). The number of included patients who underwent LLIF at our institution was 47 in the years 2006 to 2008 (group A), 155 in 2009 to 2010 (group B), and 91 in 2011 to 2012 (group C). Our data indicate a decreasing proportional trend during the past 6 years for postoperative sensory defi cits (SDs), motor defi cits (MDs), and anterior thigh pain (TP). The decreasing trends were statistically signifi cant for the proportion of

From the * Department of Orthopaedic Surgery, Spine and Scoliosis Service, Hospital for Special Surgery, Weill Cornell Medical College, New York, NY; and † Sektion für Wirbelsäulenchirurgie, Centrum für Muskuloskeletale Chirurgie, Charité-Universitätsmedizin Berlin, Berlin, Germany.

Acknowledgment date: February 11, 2013. Revision date: April 29, 2013. Acceptance date: July 1, 2013.

The manuscript submitted does not contain information about medical device(s)/drug(s).

No funds were received in support of this work.

Relevant fi nancial activities outside the submitted work: consultancy, royalties, payment for development of educational presentations, stocks.

Address correspondence and reprint requests to Alexander Aichmair, MD, Department of Orthopedic Surgery, Spine and Scoliosis Service, Hospital for Special Surgery, 535 East 70th St, New York, NY 10021; E-mail: [email protected]

Lateral lumbar interbody fusion (LLIF) via a far lateral ret-roperitoneal transpsoas approach has been established as an alternative to anterior and posterior procedures for

lumbar spinal arthrodesis. The broad spectrum of reported theoretical advantages includes the preservation of ligamen-tous structures, 1 augmentation of the intervertebral disc space with concomitant indirect decompression of neural structures within the intervertebral foramen, 2 , 3 large footprints of avail-able cages designed to span the dense apophyseal ring bilater-ally, 1 , 4 satisfactory fusion rate achievement, 5 and the avoid-ance of complications associated with traditional approaches to lumbar motion segments. 1 , 6 In fact, when compared with anterior or posterior approaches to the lumbar spine, LLIF is associated with a decreased risk of intraperitoneal visceral and vascular injuries, perioperative infections, and incidental dural tears. 6

Despite the proposed advantages of the procedure, LLIF has been reported to be associated with an increased risk of iatrogenic lumbar plexus injury as a result of dissection of psoas muscle fi bers during the far lateral approach to the spi-nal disc space. Because lumbar nerve roots converge to run

SDs in the immediate postoperative setting ( P = 0.018) and close to statistically signifi cant for SDs at last follow-up ( P = 0.126), TP immediately after surgery ( P = 0.098), and TP at last follow-up ( P = 0.136). Conclusion. To the authors’ best knowledge, this study constitutes the largest series of this sort to date, with regard to both sample size and study period. The present data indicate a decreasing proportional trend over time for SDs, MDs, and anterior TP, which can be considered a representation of an institutional learning curve during a 6-year time period of performing LLIF. Key words: lateral lumbar interbody fusion , complications , neurological defi cit , lumbar plexus , trend analysis , transpsoas approach , learning curve , sensory defi cit , motor defi cit , thigh pain , groin pain . Level of Evidence: 3 Spine 2013;38:E1483–E1490

Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

BRS205748.indd E1483BRS205748.indd E1483 03/10/13 11:11 AM03/10/13 11:11 AM

HEALTH SERVICES RESEARCH LLIF Trend Analysis • Aichmair et al

E1484 www.spinejournal.com November 2013

as a plexus within or less frequently underneath the posterior part of the psoas muscle, 7 they are prone to iatrogenic damage during the transpsoas approach in LLIF. Although intraop-erative electromyography and neuromonitoring systems have been established with the aim to reduce the risk of iatrogenic lumbar plexus injuries, adverse postoperative neurological sequelae still remain a major concern. 6 , 8 – 12

Distinct from our previously reported risk factor analyses, 13 , 14 the objective of the present study was to iden-tify the proportional trend of postoperative neurological defi -cits after LLIF over time, in a subset of patients without any neurological defi cits present before surgery. We hypothesized that the rate of neurological complications after LLIF propor-tionally decreases over time as a result of increasing surgeon experience with this procedure. To the best of our knowledge, by evaluating this trend in a cohort of 293 patients who had undergone LLIF at a single institution between 2006 and 2012, this study represents the largest series of this sort to date.

MATERIALS AND METHODS

Study Population After obtaining institutional review board approval, the elec-tronic medical records and offi ce notes of 451 patients who had consecutively undergone LLIF between March 2006 and

April 2012 at a single institution were retrospectively reviewed for reports on postoperative neurological defi cits. 13 , 14 Demo-graphic data collection included the age at index LLIF sur-gery, sex, height, weight, and body mass index. In addition, perioperative data were collected on the date of index surgery, addressed levels, side of approach, supplemental posterior instrumentation, bone graft material, and length of surgery (including the duration of general anesthesia, patient posi-tioning, and the surgical procedure itself). A “stand-alone” procedure was defi ned as the absence of supplemental pos-terior instrumentation in the setting of the index surgery. Patients with a nondegenerative indication for lumbar spinal fusion, a postoperative follow-up of less than 6 months, and/or a preoperative neurological defi cit in a distribution appro-priate to the level of the index LLIF procedure were excluded. This study was performed as part of a series of previous insti-tutional investigations focusing on LLIF. 2 , 5 , 9 , 10 , 13 , 14

Neurological Assessment Neurological data on sensory defi cits (SDs), motor defi cits (MDs), and anterior thigh pain (TP) were collected at 4 time points according to an institutional standardized protocol: in the immediate postoperative setting, at 12 weeks, at 6 months, and annually thereafter, if indicated. Assessment of neurological function by the fellowship-trained treating spine

Figure 1. A 76-year-old male patient presented to the offi ce with a long history of back and severe bilateral leg pain. Preoperative plain radio-graphs ( A ) of the lumbar spine depicted multilevel spondylosis, scoliosis, and degenerative spondylolisthesis at the L4–L5 level. On his preop-erative computed tomographic ( B , left) as well as T2 magnetic resonance imaging studies ( B , right), severe lumbar spinal stenosis was evident. Because of failure of nonsurgical treatment options, the patient underwent multilevel lateral lumbar interbody fusion at the levels L2–L3, L3–L4, and L4–L5 (Depuy Spine, Inc., Raynham, MA) via a right-sided direct lateral transpsoas approach, supplemented by bone morphogenetic protein-2 (INFUSE Bone Graft; Medtronic, Inc., Minneapolis, MN) at each level. On the same day, the patient underwent multilevel decompres-sive laminectomy (L2–L5) with medial one-third facetectomies and partial foraminotomies, as well as segmental spinal instrumentation from L2 to L5. Plain radiographs in the immediate postoperative setting ( C ) and at 2-year follow-up ( D ) depicted restored satisfactory lumbar alignment and partial reduction of the scoliotic deformity. At 2-years follow-up, the patient reported 100% improvement of his preoperative complaints.





surgeon included the evaluation of tactile detection, 2-point discrimination, and muscle strength via the manual muscle test scale (5 strength grades). In addition to an objective neurological assessment, subjective patient reports on post-operative neurological defi cits, such as sensation of pins and needles, intermittent paraesthesias/dysesthesias, pain, and/or muscle weakness, were recorded at each follow-up visit and included in the statistical analysis. An MD of (1) grade 4/5 that persisted for more than 6 months or (2) grade of 3/5 or less, independent from the duration of weakness was consid-ered a neurological MD, in contrast to a mechanical muscle injury due to psoas muscle distraction.

Surgical Technique LLIF was performed according to the surgical protocol, as similarly described in the literature. 1 In summary, after the induction of general anesthesia, the patient was secured in a lateral position to the operating table, which was then fl exed to maximize surgical accessibility of the appropriate inter-vertebral disc space. Special attention was directed toward protecting the lateral knee area to avoid compression of the common peroneal nerve. Miniopen lateral fl ank incision was followed by blunt dissection through the lateral abdominal musculature, the retroperitoneal space, and the psoas muscle to gain far lateral access to the lumbar spine. To minimize the risk for iatrogenic lumbar plexus injury, intraoperative real-time directional electromyography (NeuroVision moni-tor; NuVasive, Inc., San Diego, CA) and/or intraoperative

neuromonitoring systems were used. After transpsoas inser-tion of tubular retractors and bilateral annulotomy, the disc was removed and the intervertebral space sized with trial components. Interbody fusion cages, either from the XLIF system (XLIF; NuVasive, Inc., San Diego, CA) or from the COUGAR system (COUGAR; Depuy Spine, Inc., Raynham, MA), were implanted after graft chambers within the cages were loaded with autograft bone, allograft, or bone mor-phogenetic protein-2. Adequate cage position was confi rmed via intraoperative fl uoroscopy for each patient. If indicated, supplemental posterior instrumentation was implanted. Pre- and postoperative radiographical imaging studies of an illus-trative case are shown in Figure 1A–D .

Statistical Analysis Patients who met the inclusion criteria were classifi ed into 3 groups for statistical analysis on the basis of their LLIF sur-gery date: group A (2006–2008), group B (2009–2010), and group C (2011–2012). Continuous variables are presented as means ± standard deviations, whereas categorical variables are presented as frequencies and percentages. Time trends for the binary outcome variables of SDs, MDs, and anterior TP were analyzed using the Cochran-Armitage trend test. Because the assumption of normality was not met, the num-ber of addressed levels and length of surgery (min) by time period were compared using the Kruskal-Wallis test. All data analyses were performed using SAS software version 9.2 (SAS Institute, Inc., Cary, NC).

TABLE 1. Descriptive Characteristics of the 3 Study Subcohorts (Groups A to C)

Variable

Group A (2006–2008) Group B (2009–2010) Group C (2011–2012)

n = 47 n = 155 n = 91

Age at surgery (yr) 63.8 ± 14.1 (31–89) 61.5 ± 12.1 (29–88) 60.0 ± 11.5 (24–84)

Male sex 46.8% 36.8% 45.1%

BMI (kg/m 2 ) 26.9 ± 4.1 (16.5–36.0) 28.8 ± 5.9 (18.4–64.5) 28.5 ± 5.7 (18.3–45.7)

Level per patient 2.2 ± 1.0 (1–5) 1.9 ± 0.9 (1–4) 1.8 ± 0.8 (1–4)

LLIF levels 103 289 167

T12–L1 1.0% 0.4% 1.2%

L1–L2 11.7% 6.2% 7.8%

L2–L3 29.1% 21.8% 22.8%

L3–L4 34.0% 35.6% 35.3%

L4–L5 23.3% 35.6% 32.9%

Stand-alone LLIF 62.8% 58.7% 40.0%

BMP-2 74.5% 86.4% 87.9%

Length of surgery (min) 265.5 ± 160.8 (76–702) 257.6 ± 158.2 (69–720) 297.0 ± 171.5 (64–675)

Follow-up (mo) 21.0 ± 13.5 (6–53) 16.5 ± 8.5 (6–39) 10.5 ± 4.0 (6–20)

Continuous variables are illustrated as the mean ± standard deviation (range), whereas numerical variables are shown as frequencies or percentages.

BMI indicates body mass index; LLIF, lateral lumbar interbody fusion; BMP-2, bone morphogenetic protein-2.





RESULTS

Study Population A total of 293 patients met the study inclusion criteria and were followed postoperatively for a mean period of 15.4 ± 9.2 months (range: 6–53 mo). The study cohort pre-sented with a female-to-male ratio of 1.44:1 (173 females and 120 males), an average age at index surgery of 61.4 ± 12.3 years (range: 24–89 yr), and an average body mass index of 28.4 ± 5.6 kg/m 2 (range: 16.5–64.5 kg/m 2 ). The number of included patients who underwent LLIF at our institution was 47 in the years 2006 to 2008 (group A), 155 in 2009 to 2010 (group B), and 91 in 2011 to 2012 (group C). Detailed study demographic and perioperative data for each subcohort are shown in Table 1 .

Surgical Technique LLIF via a minimally invasive far lateral transpsoas approach was performed on a total of 559 levels by 1 of 3 fellowship-

trained senior spine surgeons (“years out of spine fellow-ship”: 17.0 ± 7.0, range: 12–25 yr). The average number of surgically addressed levels was 1.9 ± 0.9 per patient (range: 1–5 levels per patient) for the entire study population, which was also not signifi cantly different between the study sub-groups (group A: 2.2 ± 1.0 [range: 1–5], group B: 1.9 ± 0.9 [range: 1–4], group C: 1.8 ± 0.8 [range: 1–4]; P = 0.129). Overall, L3–L4 was the most frequently treated level (n = 197; 35.2%), followed by levels L4–L5 (n = 182; 32.6%), L2–L3 (n = 131; 23.4%), L1–L2 (n = 43; 7.7%), T12–L1 (n = 4; 0.7%), and T11–T12 (n = 2; 0.4%). The proportions of surgically addressed levels in groups A to C are shown in Table 1 and Figure 2 . For the entire study population, the average length of surgery was 270.5 ± 163.0 minutes (range: 64–720 min; missing data: n = 27). Subanalysis of operative times in groups A to C revealed an average length of surgery of 265.5 ± 160.8 minutes (range: 76–702 min, missing data: n = 1), 257.6 ± 158.2 minutes (range: 69–720 min, missing data: n = 13), and 297.0 ± 171.5 minutes (range: 64–675 min, missing data: n = 13), respectively. There was no statistically signifi cant difference between the 3 subgroups with regard to duration of surgery ( P = 0.219). Bone mor-phogenetic protein-2 was the most frequently used bone graft substitute overall (84.9%), as well as in each subgroup (82.9 ± 7.3%) ( Table 1 ). The postoperative neurological defi cit fi ndings for groups A to C are shown in Table 2 .

Trend Analysis: Sensory Defi cits A trend analysis showed that the proportion of patients with an SD in the immediate postoperative setting decreased across the 3 time periods (44.4% [group A], 43.4% [group B], and 25.0% [group C]; P = 0.018) ( Figure 3A ). This decreasing trend was also seen at the last follow-up time point (14.9% [group A], 11.0% [group B], and 6.6% [group C]; P = 0.126)

Figure 2. Proportions of surgically addressed motion segment in groups A to C.

TABLE 2. Proportional Trend of Neurological Defi cits of Lateral Lumbar Interbody Fusion Over Time Group A (2006–2008) Group B (2009–2010) Group C (2011–2012)

P Valuen* Defi cits † % n* Defi cits † % n* Defi cits † %

SD

Immediate postoperative 45 20 44.4 152 66 43.4 88 22 25.0 0.018

Last follow-up 47 7 14.9 155 17 11.0 91 6 6.6 0.126

MD


Last follow-up 47 2 4.3 155 4 2.6 91 2 2.2 0.523

TP


Last follow-up 47 4 8.5 155 14 9.0 91 2 2.2 0.136

*No. of patients with available data.

† No. of patients with a recorded neurological defi cit.

SD indicates sensory defi cit; MD, motor defi cit; TP, anterior thigh/groin pain. Bold value represents statistically signifi cant (P < 0.05).





( Figure 3B ). Comparing groups A and C, there was a decrease of 19.4% for an SD in the immediate postoperative setting and 8.3% at the last follow-up examination.

Trend Analysis: Motor Defi cits Immediately after the LLIF procedure, 22.2% (group A), 24.3% (group B), and 19.3% (group C) of the patients presented with an MD ( P = 0.663) ( Figure 4A ). At last follow-up, an MD was reported in 4.3% (group A), 2.6% (group B), and 2.2% (group C) of patients ( P = 0.523) ( Figure 4B ). Comparing groups A and C, there was a decrease of 2.9% for an MD in the immediate postoperative setting and 2.1% at the last follow-up examination.

Trend Analysis: Anterior Thigh Pain Anterior TP in the immediate postoperative setting was recorded in 46.7% (group A), 48.0% (group B), and 33.0% (group C) of the patients ( P = 0.098) ( Figure 5A ). Ante-rior TP at last follow-up was recorded in 8.5% (group A), 9.0% (group B), and 2.2% (group C) of cases ( P = 0.136) ( Figure 5B ). Comparing groups A and C, there was a decrease

of 13.7% for anterior TP in the immediate postoperative setting and 6.3% at the last follow-up examination.

DISCUSSION The lateral approach to lumbar motion segments via trans-psoas insertion of minimally invasive tubular retractors has the proposed advantage of avoiding intra-abdominal as well as dural injuries associated with traditional anterior and posterior interbody arthrodesis techniques. 6 This advantage, along with other reported benefi ts of minimally invasive tech-niques, has resulted in rapid popularization of LLIF among spine surgeons worldwide. However, because the lumbar plexus is situated within or less frequently underneath the pos-terior third of the psoas muscle, 7 it is prone to damage during blunt transpsoas dissection. 6 , 8 – 12 We hypothesized that the rate of neurological complications proportionally decreases over time as a result of increasing surgeon experience with this pro-cedure. To test this hypothesis, patients who had undergone LLIF between the years 2006 and 2012 were grouped accord-ing to their index surgery date, and their rates of postoperative neurological defi cits over time were compared.

Figure 3. Proportional trend analysis over time for a postoperative sensory defi cit in the immediate postoperative period ( A ) and at last follow-up ( B ) comparing groups A to C. The bar graphs illustrate the postoperative incidence of a neurological defi cit over time, as well as the associated 95% confi dence intervals.

Figure 4. Proportional trend analysis over time for a postoperative mo-tor defi cit in the immediate postoperative period ( A ) and at last follow-up ( B ) comparing groups A to C. The bar graphs illustrate the postop-erative incidence of a neurological defi cit over time, as well as the associated 95% confi dence intervals.





There is a steadily growing body of literature on the inci-dence of postoperative neurological defi cits after LLIF and potentially associated risk factors. 6 , 8 – 12 We support that the main sources for postoperative anterior thigh/groin pain, numbness, and weakness are due to approach-related injury to the genitofemoral and/or the ilioinguinal nerves as well as due to extensive psoas dissection with prolonged muscle retraction, potentially resulting in lumbosacral nerve root injury. In our previous institutional reports, besides the level of fusion and duration of surgery, the use of bone morphogenetic protein-2 as a bone graft substitute in LLIF has recently been reported to increase the risk of developing a postoperative neurological defi cit, which warrants further investigations. 9 , 13 , 14 However, to the authors’ best knowledge, there has been only 1 study to date that evaluated the infl uence of surgeon experience with this minimally invasive lumbar arthrodesis technique on neu-rological defi cit occurrence over time. In their study on lum-bar plexus injury after lumbar transpsoas interbody fusion, Le et al 15 described a proportional decrease over time potentially related to increasing surgeon expertise with LLIF. However, this study did not evaluate the degree of statistical signifi cance for the reported trends.

According to this study, the proportion of neurologi-cal defi cits at last follow-up, compared with the immediate postoperative setting, decreased by 29.6% (group A), 32.5% (group B), and 18.4% (group C), for an SD, whereas the pro-portion of MDs at the last follow-up decreased by 18.0% (group A), 21.8% (group B), and 17.1% (group C), com-pared with the immediate postoperative setting. Furthermore, at the last follow-up visit, there was a decrease of 38.2% (group A), 39.0% (group B), and 30.8% (group C) in ante-rior TP compared with the immediate postoperative setting. Our data indicate a decreasing proportional trend during the last 6 years for postoperative SDs, MDs, and anterior TP. The decreasing trends were statistically signifi cant for the propor-tion of SDs in the immediate postoperative setting ( P = 0.018) and close to statistically signifi cant for SDs at last follow-up ( P = 0.126), TP immediately after surgery ( P = 0.098), and TP at last follow-up ( P = 0.136). Furthermore, for both the immediate postoperative setting and the last follow-up exami-nation, the proportions of postoperative defi cits (SDs, MDs, and TP) were all lower in group C than in group A. This may be a result of increasing surgeon experience with the far lat-eral approach to interbody fusion, which is supporting evi-dence for an institutional learning curve for LLIF.

The effect of a learning curve in spine surgery has been previously described in the context of percutaneous arthroscopic disc surgery, 16 thoracoscopic surgery, 17 – 19 and instrumented laparoscopic spinal fusion 20 for assessing improved surgeon skills over time. 21 In fact, the infl uence of surgeon experience with a minimally invasive technique on increased surgical effi cacy is not new. In a study investi-gating the learning curve of minimally invasive transforami-nal lumbar interbody fusion, Lee et al 21 reported a gradual decrease in length of surgery over time with an asymptote after the fi rst 30 cases, as well as a decrease in intraoperative blood loss and ambulation recovery time. Similarly, a study by Neal and Rosner 22 examined the learning curve of a single neurosurgical resident between 2006 and 2008 and reported a decrease in operative time by 15.9 minutes from the initial to subsequent transforaminal lumbar interbody fusion cases, but this trend was not statistically signifi cant. Furthermore, Jhala and Mistry 23 suggested a learning phase for the initial 25 cases of minimally invasive endoscopic lumbar discec-tomy before achieving acceptable patient safety and surgi-cal effi cacy results. This is in accordance with the fi ndings of Nowitzke, 24 who estimated the learning curve asymp-tote after the performance of 30 lumbar microendoscopic discectomies.

The interpretation of the proportional trends of MDs in the postoperative setting is limited by the indistinguishabil-ity of true neurological hip fl exor MDs from pain-related weakness and/or mechanical psoas muscle fi ber injury due to muscle distraction. Another limitation is the fact that 3 spine surgeons contributed cases to this study, and in order to draw meaningful conclusions, the assumption of compa-rable learning curves was made. This is supported by the results of a previously reported multivariable binary logistic regression analysis that revealed an absence of a statistically

Figure 5. Proportional trend analysis over time for anterior thigh pain in the immediate postoperative period ( A ) and at last follow-up ( B ) comparing groups A to C. The bar graphs illustrate the postoperative incidence of a neurological defi cit over time, as well as the associated 95% confi dence intervals.





➢ Key Points

Because lumbar nerve roots converge to run as a plexus within or less frequently underneath the posterior part of the psoas muscle, they are prone to iatrogenic damage during the transpsoas ap-proach in lateral lumbar interbody fusion, and adverse postoperative neurological sequelae still remain a major concern.

Our data indicate a decreasing proportional trend during the last 6 years for postoperative sensory defi cits, motor defi cits, and anterior thigh pain.

The decreasing trends were statistically signifi -cant for the proportion of sensory defi cits in the immediate postoperative setting and close to statistically signifi cant for sensory defi cits at last follow-up ( P = 0.126), thigh pain immediately after surgery, and thigh pain at last follow-up.

The observed trends may be a result of increasing surgeon experience with the far lateral approach to interbody fusion, which is supporting evidence for an institutional learning curve for lateral lum-bar interbody fusion.

signifi cant association between one of the treating spine sur-geons and the incidence of a persistent neurological defi cit after LLIF at the last follow-up examination. 13 However, to control for the surgeon’s previous experience, we excluded 1 spine surgeon from the analysis because of the lowest number of “years out of spine fellowship” (3 yr). Never-theless, despite the inclusion of only high-volume senior spine surgeons (“years out of fellowship”: 17 ± 7.0 yr, range: 12–25 yr), who contributed cases to all study sub-groups, the decreasing proportional trends over time need to be interpreted as a learning curve of a single institution rather than that of a single surgeon. A larger sample size is needed to analyze the proportional trend separately of post-operative neurological defi cits for each individual surgeon in each individual year. Another limitation of this study is that interbody fusion implants of 2 different companies were used. Because of potential differences in the extent of psoas muscle and concomitant nerve root retraction, the incidence of postoperative neurological defi cits after LLIF may differ between the 2 used systems. However, because these implants were selected on the basis of the preference of the same surgeons who also contributed cases to all 3 study subgroups, it is unlikely that this limitation affected our results. Nevertheless, further studies are warranted to support this supposition.

With regard to the observed increase in operative length during the 6-year study time period (especially group A vs. group C), one should be aware that the recorded length of surgery includes the positioning of the patient as well as the surgical procedure itself, both being performed under general anesthesia with neuromonitoring guidance. We suggest that a potential increase in duration of patient’s positioning, which also results in an overall increase of operative time, may have been the underlying cause for our observed decreasing trend of postoperative neurological defi cits. It seems plausible that proper positioning of the fl uoroscope in addition to the achievement of a true lateral patient position may be more time-consuming but it also facilitates the surgical approach to lumbar motion segments. A larger patient series with additional differentiation between length of patient position-ing and length of surgical procedure is needed to prove this hypothesis. The surgeons now also try to reduce the duration and extent of psoas muscle retraction to minimal, yet reason-able levels in addition to being very meticulous with surgical transpsoas dissection.

CONCLUSION To the authors’ best knowledge, this study constitutes the largest series of this sort to date, with regard to both sample size and study period. The present data indicate a decreasing proportional trend over time for SDs, MDs, and TP, which can be considered a representation of an institutional learning curve during a 6-year time period of performing LLIF. Future studies investigating the learning curve for LLIF and the infl u-ence of surgeon experience on postoperative neurological complications would be benefi cial for training and furthering understanding of this spine surgical technique.

Acknowledgments The authors express their gratitude to Huong Do, MA, and Joseph Nguyen, MPH, for their assistance with the statistical analysis. They further thank Jennifer Shue, MS, for her valu-able comments during the writing of the article and her knowl-edge and insight are greatly appreciated.

References 1. Ozgur BM , Aryan HE , Pimenta L , et al. Extreme lateral interbody

fusion (XLIF): a novel surgical technique for anterior lumbar inter-body fusion . Spine J 2006 ; 6 : 435 – 43 .

2. Kepler CK , Sharma AK , Huang RC , et al. Indirect foraminal decompression after lateral transpsoas interbody fusion . J Neuro-surg Spine 2012 ; 16 : 329 – 33 .

3. Oliveira L , Marchi L , Coutinho E , et al. A radiographic assessment of the ability of the extreme lateral interbody fusion procedure to indirectly decompress the neural elements . Spine (Phila Pa 1976) 2010 ; 35 : S331 – 7 .

4. Berjano P , Lamartina C . Far lateral approaches (XLIF) in adult sco-liosis . Eur Spine J 2013 ; 22 ( suppl 2 ): S242 – 53 .

5. Kotwal S , Kawaguchi S , Lebl D , et al. Minimally invasive lat-eral lumbar interbody fusion: clinical and radiographic out-come at a minimum 2-year follow-up [published online ahead of print September 6, 2012] . J Spinal Disord Tech . doi:10.1097/BSD.0b013e3182706ce7 .

6. Rodgers WB , Gerber EJ , Patterson J . Intraoperative and early post-operative complications in extreme lateral interbody fusion: an analysis of 600 cases . Spine (Phila Pa 1976) 2011 ; 36 : 26 – 32 .

7. Kirchmair L , Lirk P , Colvin J , et al. Lumbar plexus and psoas major muscle: not always as expected . Reg Anesth Pain Med 2008 ; 33 : 109 – 14 .

8. Isaacs RE , Hyde J , Goodrich JA , et al. A prospective, nonrandom-ized, multicenter evaluation of extreme lateral interbody fusion for the treatment of adult degenerative scoliosis: perioperative outcomes and complications . Spine (Phila Pa 1976) 2010 ; 35 : S322 – 30 .

9. Pumberger M , Hughes AP , Huang RR , et al. Neurologic defi cit following lateral lumbar interbody fusion . Eur Spine J 2012 ; 21 : 1192 – 9 .





10. Sharma AK , Kepler CK , Girardi FP , et al. Lateral lumbar interbody fusion: clinical and radiographic outcomes at 1 year: a preliminary report . J Spinal Disord Tech 2011 ; 24 : 242 – 50 .

11. Sofi anos DA , Briseno MR , Abrams J , et al. Complications of the lateral transpsoas approach for lumbar interbody arthrod-esis: a case series and literature review . Clin Orthop Relat Res 2012 ; 470 : 1621 – 32 .

12. Tohmeh AG , Rodgers WB , Peterson MD . Dynamically evoked, discrete-threshold electromyography in the extreme lateral interbody fusion approach . J Neurosurg Spine 2011 ; 14 : 31 – 7 .

13. Lykissas MG , Aichmair A , Hughes AP , et al. Nerve injury after lateral lumbar interbody fusion: a review of 919 treated levels with identifi cation of risk factors . Paper presented at: 2013 Annual Meeting of the Lumbar Spine Research Society ; April 11–12, 2013; Chicago, IL . J Neurosurg . doi:10.3171/2013.4.FOC-LSRSabstracts. Available at: http://thejns.org/doi/pdf/10.3171/2013.4.FOC-LSRSabstracts . Accessed April 18, 2013.

14. Lykissas MG , Aichmair A , Hughes AP , et al. Nerve injury and recovery after lateral lumbar interbody fusion with and without bone morphogenetic protein-2 augmentation: a cohort controlled study . Paper presented at: 2013 Annual Meeting of the Lumbar Spine Research Society ; April 11–12, 2013 ; Chicago, IL . J Neuro-surg . doi:10.3171/2013.4.FOC-LSRSabstracts. Available at: http://thejns.org/doi/pdf/10.3171/2013.4.FOC-LSRSabstracts . Accessed April 18, 2013.

15. Le TV , Burkett CJ , Deukmedjian AR , et al. Postoperative lumbar plexus injury after lumbar retroperitoneal transpsoas

minimally invasive lateral interbody fusion . Spine (Phila Pa 1976) 2013 ; 38 : E13 – 20 . doi: 10.1097/BRS.0b013e318278417c.

16. Kambin P , Zhou L . History and current status of percutaneous arthroscopic disc surgery . Spine (Phila Pa 1976) 1996 ; 21 : 57S – 61S .

17. Al-Sayyad MJ , Crawford AH , Wolf RK . Early experiences with video-assisted thoracoscopic surgery: our fi rst 70 cases . Spine (Phila Pa 1976) 2004 ; 29 : 1945 – 51 ; discussion 52.

18. Lonner BS , Scharf C , Antonacci D , et al. The learning curve asso-ciated with thoracoscopic spinal instrumentation . Spine (Phila Pa 1976) 2005 ; 30 : 2835 – 40 .

19. Newton PO , Shea KG , Granlund KF . Defi ning the pediatric spi-nal thoracoscopy learning curve: sixty-fi ve consecutive cases . Spine (Phila Pa 1976) 2000 ; 25 : 1028 – 35 .

20. Zucherman JF , Zdeblick TA , Bailey SA , et al. Instrumented lapa-roscopic spinal fusion. Preliminary results . Spine (Phila Pa 1976) 1995 ; 20 : 2029 – 34 ; discussion 34–5.

21. Lee JC , Jang HD , Shin BJ . Learning curve and clinical outcomes of minimally invasive transforaminal lumbar interbody fusion: our experience in 86 consecutive cases . Spine (Phila Pa 1976) 2012 ; 37 : 1548 – 57 .

22. Neal CJ , Rosner MK . Resident learning curve for minimal-access transforaminal lumbar interbody fusion in a military training pro-gram . Neurosurg Focus 2010 ; 28 : E21 .

23. Jhala A , Mistry M . Endoscopic lumbar discectomy: experience of fi rst 100 cases . Indian J Orthop 2010 ; 44 : 184 – 90 .

24. Nowitzke AM . Assessment of the learning curve for lumbar microendoscopic discectomy . Neurosurgery 2005 ; 56 : 755 – 62 ; discussion 62.



an institutional six-year trend analysis of the

Documents