Defining the MIS-TLIF: A Systematic Review of Techniques and Technologies Used by Surgeons Worldwide

Our systematic review was performed following the guidelines proposed by the Meta-analysis of Observational Studies in Epidemiology (MOOSE) Group²⁹ and the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement.³⁰ We included articles published between 2008 and 2018. Identification of relevant studies began with an electronic search of MEDLINE. Terms included derivatives of “minimally invasive” and “lumbar spine” or “fusion.” Citations were limited to those published in English. Potential articles were then imported into the online reference management program Mendeley (Mendeley Desktop, 1.17.13, Mendeley Ltd, Elsevier, Amsterdam, Netherlands) to remove duplicate citations and organize the studies before reviewing them for study inclusion. Three independent reviewers (SL, RNH, CW) then screened study titles, abstracts, and full-text articles to identify studies reporting on MIS-TLIF. Included articles contained a detailed technical description of the MIS-TLIF and at least one of the following: (1) description of the surgical approach, (2) description of the retractor, (3) description of the interbody graft. Reviewers were not blinded in terms of the authors, journals, and/or any other data regarding the papers. A study was included in the systematic review only if all 3 screeners agreed that the study met the inclusion criteria. In cases of disagreement, the senior author (RH) was consulted regarding suitability for study inclusion. Twenty-three papers reported on the same patient cohort and used the same technique as already reported in previous publications. These papers were eliminated as they were redundant and therefore did not provide additional information. Ultimately, 75 papers were selected for inclusion (Figure 1).^{7,9-17,20-23,27,31-90}

The MIS-TLIF procedure was divided into several major components to allow a step-by-step overview and comparison of the applied techniques. The first component described was the operating room setup and equipment used, including information about intraoperative imaging and the use of computer-assisted navigation (CAN), use of the operative microscope or endoscope, and use of intraoperative neuromonitoring (IOM). This section was followed by the description of the surgical approach, including the size and location of the incision(s), the approach to the target region, and the type of retractor used. The next described component was the decompression, including the approach and details of how the decompression was performed, followed by the interbody fusion, including the type of interbody cage and types of graft and/or biologic materials used. The final component was pedicle screw and rod implantation, including the manner of insertion, unilateral or bilateral constructs, and use of Kirschner (K)-wires.

In order to perform the above-described detailed examination, data on the following features was extracted for each study: (1) name of the authors; (2) year of publication; (3) type of study; (4) total number of patients/number of MIS-TLIF treated patients; (5) diagnoses/indication for fusion; (6) levels of fusion; (7) patient positioning; (8) type of intraoperative imaging used, including use of CAN; (9) use of microscope; (10) use of IOM; (11) number of incisions; (12) location of incision(s); (13) size of incision(s); (14) type of surgical approach; (15) type and manufacturer of the retractor; (16) size of the retractor; (17) side of decompression; (18) type of decompression; (19) instruments used for the decompression; (20) timing of decompression during the procedure; (21) type of graft and/or biologics used for interbody fusion; (22) type and manufacturer of interbody cages; (23) timing of interbody fusion during the procedure; (24) use of K-wires for screw placement; (25) technique for rod placement; (26) number of screws; (27) system and manufacturer of screws/rods; and (28) timing of pedicle screw placement during procedure.

A total of 75 studies with 7808 patients (4920 treated with MIS-TLIF) were included in the analysis (Tables 1 and 2). Indications for MIS-TLIF included isthmic/degenerative spondylolisthesis, spondylolysis, spinal and/or foraminal stenosis, recurrent disc herniations, and undefined degenerative disc disease.

The majority of patients were positioned prone on a Jackson table. With the exception of one paper, all surgeries were performed under general anesthesia. Wang and Grossman applied sedation without general anesthesia.⁸⁰ Fifty-nine studies (79%) used standard fluoroscopy for intraoperative imaging, 8 studies (11%) utilized 3D fluoroscopy, and 2 studies (3%) used intraoperative computed tomography (CT); this resulted in a total of 10 publications (13%) reporting the use of CAN (Figure 2). Six studies (8%) did not mention the type of intraoperative imaging used. For visualization purposes, 25 studies (33%) specifically mentioned the use of the surgical microscope; 3 studies (4%) used an endoscope of which 2 used the endoscope through a tubular retractor and 1 study performed a percutaneous endoscopic approach; 1 study used a microscope or surgical loupes; and 1 study reported direct visualization without the use of any visualization aid. Forty-six studies (61%) did not include information about the use of the microscope or other means for enhanced visualization. With regard to IOM, 7 studies (9%) specifically reported the use of IOM, whereas the remaining 68 (91%) of studies did not mention the use of IOM.

The location of the incision(s) used for inserting the retractor was categorized into 3 groups: (1) midline requiring subperiosteal dissection, (2) close paraspinal/paramedian, and (3) a determined distance further lateral from the midline, which was further divided into (3a) in relation to a specific anatomical structure according to fluoroscopy or navigation or (3b) a measured distance from the midline (Figure 3). A single midline incision (1) was used by only 2 studies (3%), and an unspecified paramedian or paraspinal incision (2) was performed in 5 publications (7%). Three studies (4%) used a combination of a single midline incision with a paraspinal or defined distance from midline, and 1 study used either a midline or 2 paraspinal incisions. Sixteen studies (21%) made the incision(s) a specific distance from midline and the remaining 8 studies (11%) did not describe the technique for incision localization. The majority of studies (n = 40/75; 53%) used intraoperative imaging to localize the incision(s) relative to anatomic structures (3a; eg, lateral to pedicles, at the level of the facet complex). The median number of incisions was 2 (range 1-5). Almost half of the studies did not specifically mention the number or location of incisions (n = 32/75; 43%). Four publications (5%) described using a single incision, 25 (33%) used 2 incisions, 4 (5%) used 3 incisions, 3 (4%) used 4 incisions, 2 (3%) used 5 incisions, and 5 (7%) used a varying number of incisions.

The vast majority of studies utilized a type of tubular retractor to perform the decompression and interbody cage insertion (n = 61/75; 81%). Specifically, 26 publications (35%) used a nonexpandable tubular retractor, 16 (21%) used an expandable tubular retractor, 4 (5%) reported the use of either a nonexpendable or an expandable tubular retractor in their studies, while 15 (20%) reported the use of a tubular retractor but did not distinguish between nonexpendable versus expandable. Three studies (4%) used an expandable nontubular retractor, 1 study (1%) used an endoscope percutaneously without the need for a retractor, and 10 studies (13%) either did not specify or did not report the use of a retractor. The retractor size involving endoscopic cannulas, tubular, tubular expandable, as well as specular retractors varied from 8 mm to 40 mm. However, the most frequently used retractor was a 22 mm nonexpendable tubular retractor (n = 8/75; 11%). Most of the studies chose a transmuscular (n = 27/75; 36%) or Wiltse approach (n = 17/75; 23%) to reach the level of interest, and 31 studies (41%) did not specify the approach. The types of retractors used are summarized in Figure 4.

Forty-five studies (60%) performed the decompression and interbody cage insertion first (Figure 5), prior to pedicle screw placement. This decompression first group included facetectomy, nerve root decompression, discectomy, and interbody cage insertion with or without laminotomy/laminectomy prior to pedicle screw insertion. Eleven studies (15%) cannulated the pedicles and inserted K-wires as the initial step, followed by decompression and interbody cage insertion, and subsequent placement of the pedicle screw-rod constructs. Eight studies (11%) placed all the pedicle screws first. Six publications (8%) placed the contralateral screws first. Three of these studies (4%) next placed ipsilateral K-wires followed by decompression and cage insertion and lastly ipsilateral screw placement. The other three (4%) next performed the decompression and cage insertion followed by ipsilateral screw placement. One study (1%) performed the decompression first followed by, in order, contralateral screw placement, cage insertion, and ipsilateral screw placement. Last, 1 study (1%) performed the decompression first followed by bilateral screw placement and lastly the interbody cage. Only 3 studies (4%) did not report the order of steps.

When reported, the facetectomy was always performed on the symptomatic side. The majority of studies (n = 64/75; 85%) described a complete or total facetectomy, whereas 2 studies (3%) performed a partial facetectomy. The high-speed drill was used in 29 studies (39%) to perform the facetectomy and decompression with or without a pituitary rongeur, Kerrison punch, or osteotome. Twelve studies (16%) used an osteotome alone or in combination with the high-speed drill, pituitary rongeur, and/or Kerrison punches. Forty publications (53%) did not mention the instruments used for the facetectomy and decompression. When a bilateral decompression of the spinal canal was clinically indicated, 38 studies (51%) utilized a unilateral laminotomy for bilateral decompression (ULBD), 3 studies (4%) placed tubular retractors bilaterally to perform bilateral decompression, and the remaining 34 studies (45%) did not specify how bilateral decompression was performed.

Seventy-three studies (97%) reported the use of an interbody cage. One study reported the use of an “interbody graft” but did not specify exactly what was implanted and 1 study used a combination of morselized autograft, corticocancellous allograft, and demineralized bone matrix for interbody fusion. Of the 73 studies that reported the use of an interbody cage, 35 studies (48%) used a polyetheretherketone (PEEK) cage, 4 (5%) used a titanium cage, 1 (1%) used a mesh cage, 1 (1%) used either a PEEK or titanium cage, 1 (1%) used a titanium cage with a structural graft made from autologous adipose-derived stem cells, and the remaining 31 studies (42%) did not report the cage material (Figure 6). Twenty-two (30%) utilized a straight or bullet-shaped cage, 5 (7%) used a banana-shaped cage, 5 (7%) used either a straight or banana-shaped cage, 1 (1%) used a foldable cage, 1 (1%) used a mesh cage, and 39 (53%) did not mention the cage shape (Figure 7). Twenty-seven publications (37%) reported the use of a nonexpandable cage, 3 (4%) reported the use of an expandable cage, 1 (1%) used either nonexpandable or expandable, and 42 (58%) did not specify whether the cage was nonexpendable versus expandable (Figure 8). The most commonly used interbody cage was a nonexpendable, straight or bullet-shaped, PEEK cage (n = 18/73; 25%).

Again, 73 studies reported the use of an interbody cage. Of these 73 studies, 41 (56%) reported packing the interbody cage with graft material and 32 (44%) either did not pack or did not mention packing the interbody cage with graft material. Of the 41 studies that reported packing the interbody cage, 28 (68%) used autograft alone (either local autograft or iliac crest autograft) and 11 (27%) used autograft in combination with allograft, bone graft substitute, or other synthetic bone graft extender or biologic, which included 5 studies (12%) that used bone morphogenic protein (BMP) and 4 studies (10%) that used demineralized bone matrix (DBM). One study (2%) used allograft alone and 1 used autologous adipose-derived mesenchymal stem cells.

Forty-four studies (59%) reporting packing the anterior disc space with graft material and the remaining 31 (41%) studies either did not pack or did not mention packing graft into the anterior disc space. Of the 44 studies that packed the anterior disc space, the most commonly used graft material was autologous bone graft alone, either local bone or iliac crest (n = 24/44; 55%). Sixteen studies (36%) combined autograft with allograft, bone graft substitute, or other synthetic bone graft extender or biologic, which included 4 studies (9%) that used BMP, 4 studies (9%) that used DBM, 4 studies (9%) that used bone graft substitute, and 2 studies (4%) that used allograft. One study used BMP alone, 1 study used autograft or bone graft substitute, 1 study used autologous adipose-derived mesenchymal stem cells to enhance fusion, and 1 study reported the use of “bone graft” but did not specify exactly what was used. An overall summary of the graft materials utilized to enhance fusion is provided in Figures 9 and 10.

The vast majority of publications (n = 59/75; 79%) inserted pedicle screws in a percutaneous manner. Five studies (7%) used a mini-open approach utilizing expandable retractors, 2 studies (3%) directly visualized the pedicle screw entry point through an unspecified retractor, and 1 study directly visualized the entry points through a nonexpendable tubular retractor. The remaining 8 studies (11%) did not specifically mention the manner of screw placement. Forty studies (53%) performed pedicle screw placement with the use of K-wires or guide wires. The majority of studies placed rods in a percutaneous manner (n = 57/75; 76%). Seven studies (9%) placed the rods in a mini-open fashion, 2 (3%) placed the ipsilateral rod mini-open and the contralateral rod percutaneously, and the remaining 9 studies (12%) did not specify how the rods were placed. Fifty-six studies (75%) placed bilateral pedicle-screw rod constructs, 5 (7%) placed unilateral screws alone, 8 (10%) placed either unilateral or bilateral screws in their studies, 2 (3%) placed the ipsilateral rod mini-open and the contralateral rod percutaneously, and the remaining 4 (5%) did not mention placement of unilateral versus bilateral constructs. A summary of the pedicle screw insertion techniques is provided in Figures 11 and 12.

While the general steps for MIS-TLIF (decompression, discectomy and interbody cage insertion, and pedicle screw and rod insertion) are agreed upon, there is a great degree of variability in exactly how these steps are performed. To our knowledge, there is no existing review that examines the techniques reported for the performance of the MIS-TLIF, and as such, there is no accepted definition of what constitutes the MIS-TLIF. As we have demonstrated in the present review, there is great heterogeneity not only in how MIS-TLIF is performed among surgeons but also in the definition of MIS-TLIF. While many surgeons refer to the “MIS-TLIF,” it is important to realize that not all iterations of the MIS-TLIF described in the literature are truly MIS. A clear definition of the MIS-TLIF that is agreed upon by MIS surgeons is therefore needed. Such a definition would help improve clinical research as well as patient education by separating true MIS-TLIF from mini-open approaches. The benefits of the MIS-TLIF have been well reported in the literature, and one may expect these outcomes to be even better if the mini-open TLIF is removed from such studies, resulting in the comparison of true MIS-TLIF to open TLIF. Furthermore, MIS-TLIF and mini-open TLIF should be distinguished from one another so that the two can be directly compared. Is there any benefit of a true MIS-TLIF over a mini-open TLIF? Or is there a degree of invasiveness below which there is no additional benefit? Additionally, a clear definition of MIS-TLIF is required to better inform patients. The “MIS-TLIFs” offered to patients by different surgeons, as we have demonstrated in this systematic review, have a great degree of variability in the degree of invasiveness. One surgeon’s “MIS-TLIF” may rather represent a mini-open approach compared with another surgeon’s true MIS-TLIF. The aim of the present systematic review was to examine the various MIS-TLIF techniques reported in the existing body of literature and to identify patterns that represent accepted techniques among spine surgeons to aid in defining the MIS-TLIF. We also aimed to identify the techniques that require a higher degree of exposure and thus increased tissue damage that should disqualify a TLIF from being labeled as MIS, and in doing so identify several components or techniques of the MIS-TLIF that contribute to the degree of invasiveness (Figure 13). Our review has illuminated several such patterns.

With regard to the agreed upon components of the MIS-TLIF, the majority of studies (79%) used standard fluoroscopy as the means of intraoperative imaging, with only a small minority using advanced imaging such as 3D fluoroscopy or intraoperative CT with CAN. This was an interesting finding as in our practice we almost exclusively use intraoperative CT with CAN,⁵⁷ and the field of MISS, as well as patient interest, seems to be moving toward more emphasis on navigation. Additionally, multiple studies have demonstrated the increased accuracy of pedicle screw placement with CAN relative to fluoroscopy,^92-96 and a meta-analysis demonstrated less blood loss and fewer complications with the use of CAN.⁹⁷ The cost of purchasing such equipment is certainly a burden, particularly in smaller or community hospitals. Perhaps as the technology continues to improve and become more available and affordable, its utilization may increase in practice with more surgeons adopting these advanced intraoperative imaging systems.

As may be expected for an MIS procedure that requires percutaneous pedicle screw placement, over 90% of studies reported the use of incisions off midline, most commonly in relation to a specific anatomic structure (eg, the pedicle or facet joint). The incision location also allowed the vast majority of surgeons to apply either a muscle sparing Wiltse approach or a less specific transmuscular approach to reach the level of surgical interest. Both approaches fit the key principle of MISS of minimizing muscle and soft tissue disruption. The majority (79%) of studies reported the placement of pedicle screws in a percutaneous manner and 75% placed bilateral pedicle-screw rod constructs.

Over 80% of publications used a tubular retractor, either expandable or nonexpandable. A small minority reported the use of an expandable nontubular retractor such as a blade retractor that requires extensive superiosteal dissection. Fourth, when reported, almost all studies performed a total facetectomy on the symptomatic side, and when a bilateral decompression was deemed necessary, 93% of publications that described the details of achieving bilateral decompression did so via ULBD.

Last, all studies except two reported the use of an interbody cage. When graft material was used, 95% and 91% used autograft with or without allograft or bone graft substitute to pack the anterior disc space and interbody cage, respectively. The use of different materials for performing the interbody fusion has been previously described by our group, and overall, fusion rates for MIS-TLIF are high regardless of the graft materials used.⁵⁷ In our study, only 16% of the authors reported the use of BMP alone or in combination with any other materials to enhance fusion. In previous studies focused on fusion materials for MIS TLIF procedures, higher prevalence of BMP use reported has been reported.⁹⁸ However, our inclusion criteria were focused on technical nuances rather than details about fusion grafts. Therefore, it is possible that the prevalence of surgeons using BMP is underestimated in our article.

To summarize, the existing body of literature describes the MIS-TLIF to entail the use of paramedian incisions to perform decompression, total facetectomy, and interbody cage placement via a tubular retractor. When graft material is used, autograft is the gold standard. If bilateral decompression is indicated, ULBD is utilized. Finally, the pedicle screw-rod constructs are placed in a percutaneous manner with the use of intraoperative imaging, most commonly standard fluoroscopy.

With regard to workflow, most surgeons performed the decompression and interbody cage insertion prior to pedicle screw placement. When pedicle screws were placed first, many authors reported temporarily distracting on the screws in order to open the disc space, followed by compression once the interbody graft was placed. Of note, some of the mini-open retractor systems use the pedicle screw heads and towers for retraction. When using CAN, the pedicle screws or K-wires should be inserted first as insertion of an interbody cage can change the location of the pedicle in anatomical space and result in screw misplacement if an intraoperative scan is not obtained after interbody placement. For this reason, in our practice, we obtain a CT at the start of the procedure and place the pedicle screws first, followed by decompression and interbody cage insertion so as to expose the patient to a single radiation dose at the start of the procedure. Therefore, the order in which the basic steps of the TLIF are performed is not necessarily an indicator of the degree of invasiveness. Other factors that did not contribute to the level of invasiveness included the type of intraoperative imaging used, the use of IOM, the type of interbody cage used, and the type of graft material used.

Examination of our systematic review also yielded components of the TLIF that contributed to the degree of invasiveness and certain components that had no effect on the degree of invasiveness. Importantly, we have identified several techniques that we believe should disqualify a TLIF from being considered truly MIS due to wider exposure and subsequently increased tissue trauma. While there are multiple factors contributing to invasiveness, we managed to narrow criteria for MIS-TLIF down to 3 main criteria.

First, perhaps the most critical factor to the MIS-TLIF definition elucidated in the present review is the type of retractor used. As the vast majority of publications reported the use of tubular retractors (81%), the use of a nontubular expandable retractor should disqualify a TLIF from being defined as MIS. Expandable nontubular retractors with subperiosteal dissection rather represent a mini-open approach, and while this technique remains less invasive than the traditional open TLIF, it represents a more invasive TLIF than that obtained with a nonexpendable tubular retractor. Along the same lines, the use of an expandable tubular retractor represents a TLIF that may be more invasive than that achieved with a nonexpandable tubular retractor. If a “pedicle-to-pedicle” exposure is achieved via an expandable tubular retractor, one must not consider this an MIS-TLIF, but rather, again, a mini-open variant. Based upon the type of retractor used, we have identified 5 tiers of invasiveness: (1) most invasive traditional open TLIF; (2) mini-open approach with the use of an expandable nontubular retractor; (3) use of an expandable tubular retractor; (4) use of a nonexpendable tubular retractor; and (5) least invasive with the use of a percutaneous endoscopic approach (Figure 13).

Second, the use of a midline incision should disqualify a TLIF from being described as MIS. Assuming the midline incision is used to perform a subperiosteal dissection and to achieve lateral exposure for the correct bilateral pedicle screw trajectory, it would require a large incision not true to the spirit of MIS.

Moreover, the use of some kind of visualization aid such as surgical loupes, a surgical microscope or endoscope is required for the MIS-TLIF (Figure 13). The access corridors granted by the tubular retractors used in MIS-TLIF are narrow and visualization without magnification or additional illumination is poor. If a microscope or endoscope is not utilized, one must question the MIS nature of the procedure. Within the included articles, only 61% explicitly reported the use of a visualization device. However, since the majority of authors reported the use of a tubular retractor, we believe it is safe to assume that these procedures were performed with some sort of magnification aid. Overall, these criteria for MIS-TLIF were met by 81% of the included articles while 12% were identified as mini-open approach and 7% rather described an open approach per our above-mentioned criteria. Furthermore, we could not identify any specific geographical preferences or differences among the articles reporting about MIS-TLIF.

There are several limitations to this review of the literature: (1) the review was conducted retrospectively and mostly reflects state of the art practice of the past 8 years; (2) we included studies independent of their quality and level of evidence; and (3) inclusion criteria were broad. Nevertheless, these limitations may not have much impact on our findings and conclusion, as the aim was to identify some degree of consensus described in the literature for the performance of MIS-TLIF and not to report the outcomes of such techniques. We also relied on detailed descriptions of the MIS-TLIF by the authors to synthesize our data. There were many aspects of the MIS-TLIF that were not specifically described or overlooked by the authors. For example, 61% of studies did not mention the use of the microscope, endoscope, or other means for enhanced visualization. However, it is unlikely that these studies did not use any equipment to improve magnification and illumination. An additional limitation relates to the perpetual advancement of MISS. With the improvements and advancements in TLIF techniques and equipment, the manner in which the procedure is performed must adapt. This is clearly demonstrated if one compares the evolution and adoption of the MIS-TLIF over the past 2 decades. Therefore, one may expect that newer publications represent less invasive techniques than older publications. We have attempted to limit this bias by examining studies published only within the past 10 years. Nevertheless, a surgeon who published on his or her series 5 years ago utilizing a mini-open approach may now have transitioned to a truly MIS approach but has not published on his or her new series; this change in practice would not be captured in the present review. As surgeons are exposed to MIS techniques earlier in their training as residents or fellows and therefore more comfortable with MIS approaches upon completion of training, the number of surgeons performing MIS techniques can be expected to increase in the coming years. This is yet another reason why a clear definition of MIS-TLIF is needed among minimally invasive spine surgeons.