Lowe, Thomas G.*; Tahernia, A. David†; O'Brien, Michael F.*; Smith, David A. B.*

INTRODUCTION

Blume, in 1981, described a unilateral approach to posterior lumbar interbody fusion (PLIF) to address some of the potential complications of the standard PLIF ^(1,25). The technique has been recently popularized by Harms et al. ⁽²⁾. The unilateral transforaminal posterior lumbar interbody fusion (TLIF) is a surgical technique in which bilateral anterior column support can be achieved through a unilateral posterolateral approach. Posterior as well as anterior column stability is achieved when pedicle screw fixation is added.

To our knowledge, there have been no published reports assessing the outcome of patients treated with a TLIF procedure for degenerative disorders of the lumbar spine. We prospectively followed up 40 patients treated with this technique and report on the clinical and radiographic results with a minimum 2-year follow-up.

TECHNIQUE

The patient is placed on a suitable spine frame in the prone position with the hips in maximum extension, helping to maintain lumbar lordosis and when present affording partial reduction of an isthmic or degenerative spondylolisthesis. A standard midline approach is used. Careful dissection is performed out to the tips of the transverse processes of the levels included in the fusion. Through the same incision one iliac crest is exposed, but muscles are left attached. A small window is made in the iliac crest and an adequate amount of cancellous bone is removed from between the cortical tables of the iliac crest to use for bone graft. The fascia over the iliac crest is closed with interrupted sutures.

A localizing radiograph is taken for level identification purposes. Multiaxial pedicle screws are inserted at the appropriate levels, and proper placement is confirmed with biplanar fluoroscopy and direct stimulation of the screws while recording electromyograph responses in the adjacent nerve roots (evoked electromyographs). Somatosensory evoked potentials and both spontaneous and evoked electromyography provide feedback on nerve root function, especially during insertion of pedicle screws and distraction of the disc space ^(3–5).

The side of the spine selected for the TLIF is based on preoperative symptoms. If a disc herniation or foraminal stenosis is present and predominantly one-sided, then that side is chosen. Central decompression, if indicated, is then performed based on preoperative symptoms and imaging studies. A partial facetectomy is carried out on the contralateral side. After bilateral pedicle screw insertion, the rods are contoured in slight lordosis and cut approximately 1 cm longer than would be normally used to allow for disc space distraction. The rod and locking plugs are inserted into the multiaxial screw heads and as bilateral distraction is applied, the locking plugs are tightened. Multiaxial screws are recommended by the authors because they allow for concentric seating of the rods when distractive and compressive forces are applied during the procedure. A 0.25-inch osteotome and Kerrison rongeurs are used to remove the entire inferior articular process and superior portion of the superior articular process on the side chosen for the TLIF procedure.

Exposure of the underlying disc space is facilitated by removal of ligamentum flavum and underlying fatty tissue but preserving fatty tissue surrounding the nerve root. Identifying the exiting nerve root inferior to the upper pedicle to be instrumented helps orient the surgeon, because the remainder of the anatomy is consistent in relation to this structure. Epidural bleeding is frequently encountered at this point. Irrigating bipolar electrocautery is very useful in controlling epidural bleeding, and thrombin-soaked absorbable gelatin sponges and cottonoids are also used if needed. Once hemostasis is achieved, the underlying disc space (lateral one-third), dural sac, and exiting nerve root should be readily visible. The exiting nerve root rarely needs retracting except at the L5–S1 level. A 45° nerve root retractor placed around the dural sac is used to protect that structure.

A 15-blade scalpel is used to create a rectangular annular window. The medial border of the window is the lateral margin of the dural sac, and the lateral border is the lateral edge of the visible annulus. The annular window is caudal to the exiting nerve root, and because the approach is transforaminal, the exiting nerve root rarely requires retraction during the procedure. The incised annulus is removed with a pituitary rongeur. The first cage is inserted into the annular window and rolled to the contralateral side of the disc space using angled impactors. The previously inserted packed morselized graft facilitates placement of the cages in the posterior half of the disc space. The second cage is then inserted through the same annular window and positioned in the ipsilateral side of the disc space with a straight impactor. After insertion of cages proper positioning is documented with biplanar fluoroscopy. Care should be taken not to place the cages in the anterior half of the disc space because of the risk of creating foraminal stenosis when compression is applied to the pedicle screws later in the procedure. A 0.25-inch osteotome is used to enlarge the window and remove posterior osteophytes, allowing easy access to the disc space. Specialized straight and angled osteotomes, pituitary rongeurs, rasps, and curettes are used to elevate and remove disc material. Additional distraction is applied to the pedicle instrumentation periodically during the procedure for additional disc space visualization when needed, relying on ligamentotaxis. The disc space is irrigated with bacitracin-containing saline and then reinspected to confirm complete removal of disc material. A disc spanner is inserted through the annular window to determine the appropriate size cage, which should be approximately 1 mm smaller in height than the disc spanner measurement to allow for lordosis when compression is applied to the posterior pedicular instrumentation; 13- or 14-mm-diameter cages with heights ranging from 7 to 9 mm are most commonly used. A 0.25-inch angled osteotome is used to decorticate only the anterior half of the adjacent endplates. This endplate removal and decortication provides an excellent graft bed adjacent to the anterior annulus. The posterior half of the adjacent endplates are left intact to provide support for the cages. Previously harvested iliac crest bone graft mixed with local bone is then tightly packed into the anterior disc space with a bone tamp. Two titanium cages are then packed with autograft and inserted into the disc space as shown in Figure 1. The first cage is inserted into the posterior interbody interspace and maneuvered across the disc space to the contralateral side using straight and angled impactors. The second cage is inserted into the ipsilateral posterior disc space. Pedicle screw distraction is then released, and biplanar fluoroscopy is used to confirm proper placement of the cages. This provides structural support close to the axis of rotation for the motion segment maximizing lordosis when compression is applied to the pedicle screws and allows later radiographic assessment of fusion anterior to the cages. Compression is then applied, locking the cages in place and facilitating segmental lordosis. The contralateral facet and bilateral transverse processes are decorticated and packed with iliac bone graft as shown in Figure 2A and B. The wound is closed in layers over closed-suction drainage completing the procedure.

Fig. 1

Fig. 2

Patients are mobilized on postoperative day 1, and no external orthosis is required. For the first 6 weeks postoperatively, patients are instructed in a progressive walking program. At 6 weeks, progressive range of motion and strengthening exercises are then initiated and by 3 months patients are allowed low impact activities as tolerated. Full activities are resumed at 6 months. Patients were then followed at regular intervals with periodic radiographs until latest follow-up (>24 months).

METHODS

Patients

Forty patients (age range: 24.1–69.2 years) underwent a transforaminal interbody fusion (TLIF) from June 1996 to November 1998; all of them had more than 6 months of disabling back pain with or without leg pain refractory to aggressive conservative treatment. There were 17 women and 23 men. Five patients were significant smokers (>1 pack/day) and six had worker's compensation back injuries.

Radiographic Assessment

Plain posteroanterior and lateral standing radiographs including flexion–extension lateral views were obtained to evaluate disc height, segmental instability, spondylolisthesis, sagittal profile, and balance. Magnetic resonance imaging (MRI) scans were obtained in each patient to document levels of degenerative disc disease and sites of neural compression. In cases when radiographic evidence of spondylolisthesis with translational instability was not present, provocative discography was used to confirm or reject suspected symptomatic degenerative levels noted on MRI scan. A discogram was considered positive when a radial tear was present with a minimum of 8/10 concordant pain in the presence of a normal adjacent discogram. Postoperatively plain radiographs including flexion–extension views were obtained at regular intervals to assess the progress of the fusion. A fusion was confirmed by progressive increase in interspace bone density and blurring of adjacent endplates, presence of bridging posterolateral trabecular bone, and no evidence of hardware failure, loosening, or motion on flexion–extension radiographs ⁽⁶⁾. Case 1 (Figure 3A–G) demonstrates an excellent fusion mass bilaterally at 24 months postoperatively in a patient with degenerative spondylolisthesis at L4–L5.

Fig. 3

Clinical Outcome

Patients were asked to complete pre- and postoperative questionnaires assessing pain (medication use) and ability to perform activities of daily living (ADLs) including lifting, walking, standing, sitting, work status, and social activities. The questionnaire was largely based on the Oswestry outcomes instrument. A point system (Table 1) with a maximum value of 34 was used to categorize results as excellent, good, fair, unchanged, and poor. Previous lumbar surgery, smoking history, worker's compensation, and disability status were also recorded. Patients were also asked if they would have the surgery again based on the degree of perceived improvement or deterioration in pain and function from their preoperative status.

Table 1

RESULTS

There were 23 women and 17 men with a mean age of 44.9 years (range: 24.1–69.2 years) who underwent a TLIF from June 1996 to November 1998. Follow-up averaged 36 months (range: 31– 42 months). Thirty-four patients underwent a single-level TLIF, and six had two levels fused. All 40 patients had degenerative disc disease and in addition, 13 had grade I to III isthmic or degenerative spondylolisthesis and 4 had a recurrent disc herniation at the L4–L5 level. Radiographic fusion was thought to be present in 36 of 40 (90%) of the patients based on the presence of obliteration of the disc space anterior to the cages as well as continuous trabecular bone throughout the intertransverse fusion mass, no loosening or breakage of implants, and no demonstrable motion on flexion–extension radiographs. One patient had a confirmed pseudoarthrosis requiring reoperation, and three had radiographic evidence suggestive of pseudoarthrosis based on 2-year radiographs but were not symptomatic to a level of considering further surgery. This yielded a confirmed pseudoarthrosis rate of 1 of 40 or 2.5% and a total suspected pseudoarthrosis rate of 4 of 40 or 10%. Segmental lordosis for one-level and two-level TLIFs improved 29.6% and 13.6%, respectively, over preoperative levels.

Questionnaires were filled out by all 40 patients. Pain level on a 10-point visual analog scale improved from a preoperative mean value of 8.3 ± 1.97 to 3.2 ± 2.06 (paired t test, p < 0.0001) at latest follow-up (Fig. 4). No patients reported postoperative pain greater than their preoperative level. Preoperatively, 36 of 40 patients were taking at least one oral narcotic analgesic daily for pain, and postoperatively, four patients were still taking oral narcotics at latest follow-up and 10 patients were taking nonsteroidal antiinflammatory drugs for pain control. The preoperative composite score for activities of daily living (Oswestry) preoperatively averaged 9 of 24 ± 3.1 and postoperatively increased to 16 of 24 ± 2.3 (p < 0.001) (Fig. 5). No patients in the series became less able to perform ADLs postoperatively. When asked if they would have the surgery again, based on their outcome, 33 patients (82.5%) said they would. Similarly, 80% (32/40) of the patients were rated excellent or good based on pre- and postoperative questionnaire scores that included combined pain and ADL scores. There were six fair and no unchanged or poor results based on questionnaire scores.

Fig. 4

Fig. 5

Preoperatively, there were 23 patients working and 10 were not working because of back pain, and 7 were retired. Postoperatively, all 23 patients who were working returned to work, 1 retired, and 8 of 10 who were not working because of back pain returned to light work postoperatively. All of the patients who were not working preoperatively remained so postoperatively.

There were no statistically significant differences in fusion rates between smokers and nonsmokers. Likewise, there were no statistically significant differences in clinical outcome between worker's compensation and non–worker's compensation patients in this series, although there were only five smokers and six worker's compensation patients in the series.

Complications other than pseudoarthrosis (4/40) included two dural tears repaired intraoperatively, one transient neurapraxia, and one late infection (>2 years) necessitating removal of pedicle instrumentation. The fusion was noted to be solid during hardware removal in that patient. There was no evidence of clinical arachnoiditis or cage-related complications in any of our patients.

DISCUSSION

Since the first reports of spinal arthrodesis 88 years ago, a variety of techniques for fusion of the lumbar spine have been developed for the management of a wide range of conditions ^(7,8). The rates of spinal fusion with bone graft alone have ranged from 46% to 90%^(9–12,27). Because of difficulty in achieving fusion and maintaining spinal alignment and position, spinal instrumentation has become an important and popular adjunct to bone grafting in lumbar arthrodesis surgery, further increasing the fusion rates (80–90%) ⁽¹³⁾.

More recently, interbody fusion techniques have also shown high fusion rates with distinct advantages ^(14–17,26). Some of the advantages include immediate anterior column load sharing, a large surface area for fusion, bone graft subjected to compressive loads that is advantageous in achieving fusion, and the ability to restore normal sagittal contour while indirectly decompressing the neuroforamen ⁽¹⁴⁾. Interbody fusion techniques also appear to be the most effective treatment of discogenic back pain that is unresponsive to conservative care ^(18,19). Weatherly et al. reported on five patients during a 10-year period who had “solid” posterolateral fusions but still had positive discography under the fusion and had their back pain relieved by anterior interbody fusions ⁽¹⁹⁾. All five patients had positive discograms and had pain relief after anterior interbody fusion. Recently, Derby et al. noted that patients with “highly sensitive discs” as determined by pressure-controlled discography achieved significantly better long-term outcomes with combined anterior/posterior fusion than with intertransverse fusion alone ⁽¹⁸⁾.

Interbody fusion can be achieved through either an anterior, posterior, or combined approach. Proponents of the anterior approach cite shorter hospital stays and the prevention of “fusion disease,” which is thought to occur secondary to paraspinal muscle dissection during posterior exposure for fusion ^(20,21). High fusion rates and patient satisfaction have been reported with this approach ⁽²²⁾. However, complications do occur related to improper cage placement and exposure (injury to iliac vessels, retrograde ejaculation, etc.). The efficacy of threaded cage devices as “stand alone” implants for treatment of the degenerative lumbar spine at the present time are probably not indicated for other than the “collapsed disc”⁽²¹⁾.

Posterior interbody techniques allow the surgeon to simultaneously address all of the pathological lesions through a single approach. Shorter incisions and care in muscle stripping have resulted in less soft-tissue dissection. When combined with pedicle screw fixation, anterior and posterior column stabilization can be achieved. The addition of an interbody fusion to a posterolateral fusion provides a 360° circumferential fusion bed and may be associated with improved fusion rates, especially in patients with other medical comorbidities (i.e., diabetes, obesity, and nicotine abuse) who are at greater risk of pseudoarthrosis ^(13,23,24). Biomechanical studies of posterior lumbar interbody fusion without additional posterior instrumentation have suggested that significant destabilization of the fused segment may occur ⁽²³⁾. Significant bilateral bony and ligamentous removal is often required to allow accurate placement of properly sized implants, and it is not possible to provide segmental lordosis unless posterior instrumentation is added. Perhaps the greatest concern with a standard PLIF is the amount of neural retraction needed, potentially leading to nerve root injury, dural laceration, and epidural fibrosis. Cauda equina injuries occurred in 19% of patients in one series with permanent nerve dysfunction in three patients ⁽²⁵⁾. Ray noted 13 dural tears and a 10% incidence of transient foot weakness in his follow-up study of 236 patients treated with posteriorly inserted threaded cage device ⁽¹⁶⁾.

The transforaminal unilateral posterior lumbar interbody fusion was developed to address some of these problems. Advantages over the standard PLIF include the ability to provide bilateral anterior column support through a single posterolateral approach of the disc space. Because of the transforaminal approach, this technique preserves the anterior and most of the posterior longitudinal ligamentous complex, which provides a tension band for compression of the graft and prevents retropulsion of the graft. It avoids excessive soft tissue dissection, which may help prevent scarring and instability of adjacent segments, as well as injury to the exiting nerve root. Epidural bleeding is less of a problem than with the standard bilateral PLIF because of the unilateral transforaminal approach, and with experience, proper cage placement within the disc space is consistently achieved.

The TLIF is indicated for anterior column deficiency with chronic mechanical pain related to degenerative disc disease, recurrent disc herniation, and/or spondylolisthesis. Segmental kyphosis related to disc narrowing can be restored to a relatively normal lordotic contour, theoretically decreasing pathologic forces on adjacent motion segments, which may be a factor leading to adjacent segment degeneration or “transitional syndrome.” Grade II or III spondylolisthesis can also be reduced with this technique without the need for anterior surgery. Theoretically, there should be a higher rate of arthrodesis with the TLIF because the anterior graft is loaded in compression. Patients with multiple comorbidities potentially affecting successful arthrodesis (obesity, smoking, diabetes, previous failed fusion) are also candidates for the TLIF because it provides a circumferential fusion through a posterior-only approach. The TLIF procedure can also be used to decrease pseudoarthrosis at the lumbosacral junction in cases when long fusion to the sacrum is required such as in de novo degenerative scoliosis, where traditionally a combined anterior–posterior fusion would be required.

Our radiographic fusion rate of 90% and objective clinical excellent/good rate of 85% compare favorably with previous reports using other fusion techniques. This may be partly because of an overall favorable patient population. The majority of our patients were not heavy smokers nor were they seeking workers' compensation disability at the time of surgery. However, our results partly underscore the importance of proper patient selection when considering surgery to treat degenerative diseases of the lumbar spine.

The point system used in our questionnaire has not been entirely validated. However, using the same format in our pre- and postoperative questionnaires based on the Oswestry outcome instrument maintained consistency when comparing preoperative disability to 2-year follow-up clinical outcome. We believe that the assessment of pain and the ability to perform ADLs at latest follow-up enabled us to determine clinical success in our patient population and is consistent with the parameters used to determine clinical outcomes in previous studies.

Our study is not a comparative study. Our outcomes can be compared with similar studies in the literature for other fusion techniques, but no control group was included in our study. Our goal was merely to present our early results with this procedure and to offer it as a reasonable alternative for appropriately selected patients.

The TLIF can be easily mastered, but a learning curve exists. We have experienced no difficulties with implant failure related to the use of pedicle instrumentation as a means of distraction. The exiting nerve root must be visualized throughout the procedure. Meticulous attention to disc removal is essential. This allows the largest surface area possible for fusion and aids in proper cage placement. We have found that incomplete removal of disc material just ventral to the posterior annulus can compromise cage placement. Because a unilateral approach is used, the surgeon is relying on indirect visualization and tactile feedback when working across the disc space. Specialized instruments greatly facilitate disc removal during this portion of the case. We do not recommend this procedure for patients with severe osteopenia (bone mineral density <60% predicted). Other contraindications include bilateral epidural fibrosis pseudoarthrosis or a fusion of more than two levels.

CONCLUSION

The transforaminal posterior lumbar interbody fusion is a safe and reproducible technique to provide bilateral anterior and posterior column support through a unilateral posterior approach. The ideal patient for this procedure is one with long-standing mechanical back pain with or without a significant radicular component unresponsive to aggressive nonoperative treatment with radiographic evidence of a deficient anterior column. A sound biomechanical construct is achieved with a large surface area for achieving a successful fusion. This technique offers the additional advantage of providing for improved segmental lordosis and reduction of isthmic or degenerative spondylolisthesis. High fusion rates with good clinical outcomes can be achieved with few complications using this technique. Proper patient selection continues to be the most important factor in good clinical outcome with this procedure as well as others.

REFERENCES