Keywords discectomy, percutaneous - disc herniation - iliac crest - minimally invasive surgical procedures
Introduction
The Kambin triangle is the base for transforaminal endoscopic surgery.[1 ] This anatomical structure is delimited by the superior vertebral plateau, the dural sac, and the emerging root. The space within this triangle represents a safety corridor free of noble structures. The introduction of instruments small enough to enter this space allowed the development of the transforaminal (TF) approach.[2 ]
[3 ] In this approach, the working cannula aligns with the intervertebral disc, crossing the safety triangle with minimal injury and joint preservation, allowing early recovery and greater joint stability.[4 ]
However, the TF approach at the L5-S1 level is unique due to the iliac crest since the cannula introduction occurs superior to the intervertebral disc to avoid obstructions by the iliac crest, resulting in a more angled alignment about the disc, called the suprailiac (SI) approach. When the iliac crest is prominent, the SI approach may be unfeasible, requiring crest perforation, that is, a transiliac (TI) approach. Osman and Marsolais[5 ] validated the TI approach safety in cadaveric studies, but concerns regarding intraoperative injuries, bleeding, and pain remain.
The TF approach has benefits, as it allows foraminal decompression, avoids root manipulation, and can occur under sedation.[6 ] Choi et al.[7 ] postulated that patients whose lateral spine radiographs demonstrated an iliac crest above the lower half of the L5 pedicle tend to face significant challenges in the TF approach. Patgaonkar et al.[8 ] proposed a classification outlining when to select an SI or TI approach.
The present study aimed to evaluate lumbar spine radiographs using the Choi[7 ] and Patgaonkar et al.[8 ] classifications to determine the parameters potentially influencing the L5–S1 TF approach and verify the presence of mega-apophysis in the evaluated sample.
Materials and Methods
We performed a cross-sectional study using lumbar spine radiographs from our institution, which we collected over 4 months. We included patients over 18 years old, with radiographs with adequate visualization and excluded those with previous lumbar spine surgery, tumor lesions, fractures, and scoliosis.
We evaluated the radiographs using the Myvue system, version 11.2.2.3 (Carestream Health, Rochester, NY, USA). On the frontal radiograph, we measured the iliac crest height (ICH), that is, the vertical distance between a line tangentially connecting the tops of the iliac crests and the superomedial edge of the S1 joint, and the iliac rim angle (IRA), measured at the intersection of the horizontal line with the line passing through the superomedial edge of the S1 joint and tangentially connecting the medial edge of the iliac bone ([Fig. 1 ]). The pelvis classification as android or gynecoid occurred based on its morphology, and the mega-apophysis classification followed the aspects proposed by Castellvi apud Konin and Walz.[9 ]
Fig. 1 Iliac crest height (ICH; double arrow): distance between the superomedial edge of the L5–S1 vertebral joint and the line tangential to the top of the iliac crests. Iliac rim angle (IRA; dotted line): the angle between the horizontal and the line passing through the superomedial edge of the L5–S1 joint and tangential to the medial edge of the iliac crest.
The classification by Patgaonkar et al.,[8 ] used in anteroposterior radiographs, evaluates the relationship of the L5 pedicle with a line drawn from the top of the iliac crest to the center of the lower plateau of the L5 vertebra. In type I, the line is below the pedicle; in type II, the line passes tangent to the lower edge of the pedicle; and, in type III, the line crosses the pedicle ([Fig. 2 ]). According to Patgaonkar et al.,[8 ] patients classified as type III are suitable for a TI approach. For types I and II, the indication is an SI approach, and, for type III, a TI approach.
Fig. 2 Anteroposterior Patgaonkar classification. Type I: the line between the highest point of the iliac crest and the center of the lower plateau of L5 passing below the L5 pedicle; type II: the same line passing tangentially to the lower edge of the L5 pedicle; type III: the same line passing through the L5 pedicle.
This same classification assesses lateral radiographs using the top of the iliac crest and the upper and lower edges of the L5 pedicle. In type I, the iliac crest is below the pedicle; in type II, it is at the level of the pedicle; and, in type III, it is above the pedicle ([Fig. 3 ]). As in the previous classification, the author[8 ] considers type-III patients eligible for the TI approach. For types I and II, the indication is a single SI approach, and, for type-III cases, a TI approach.
Fig. 3 Lateral Patgaonkar classification. Type 1: the line between the highest point of the iliac crest and the center of the lower plateau of L5 passing below the pedicle of L5; type 2 the same line passing tangentially to the lower edge of the pedicle of L5; type 3: the same line passing through the pedicle of L5.
Choi et al.[7 ] presented a similar classification, defining the height of the iliac crest by lumbar spine structures. Types 1, 2, and 3 were considered the easiest to access without the need for foraminoplasty and grouped as suitable for SI approach. In this classification, types 5 and 6 are above half of the L5 pedicle and associated with greater difficulty for the TF approach, often requiring a foraminoplasty; for these types, the indication is the TS approach with foraminoplasty. The addition of type 7 occurred only to classify iliac crests above the L4-L5 intervertebral disc, which were not foreseen by Choi et al.[7 ] ([Fig. 4 ]).
Fig. 4 Choi classification - the diagram represents a lateral lumbar spine radiograph. Choi defined the first six reference types for the iliac crest height. We added the seventh type to include types not foreseen in the classification. The dotted curved line represents the overlap of the iliac crest on the radiograph.
We described the quantitative characteristics evaluated according to the approach and iliac type for the Patgaonkar classification,[8 ] and compared with the categories using Student's t-tests, and per the Choi classifications,[7 ] using analysis of variance (ANOVA).[10 ] We described gender and iliac type per the approaches from each classification and verified associations using likelihood ratio tests.[11 ]
The Spearman's test calculated the correlations between ICH and IRA, illustrated as scatter diagrams. We performed the analyses in IBM SPSS Statistics for Windows, version 22.0 (IBM Corp., Armonk, NY, USA) and tabulated the data on Microsoft Excel 2013 (Microsoft Corp., Redmond, WA, USA). The significance level was 5%.
Results
We obtained a total of 167 radiographs and described their characteristics in [Table 1 ]. [Table 2 ] describes their classifications according to Patgaonkar et al.[8 ] and Choi et al.[7 ]
Table 1
Variable
Description
(N = 167)
Age (years)
Mean ± standard deviation
49.1 ± 16.7
Median (minimum–maximum)
48 (18–87)
Gender, n (%)
Female
82 (49.1)
Male
85 (50.9)
Iliac type, n (%)
Android
75 (44.9)
Gynecoid
92 (55.1)
Iliac crest height (cm)
Mean ± standard deviation
25.9 ± 7.5
Median (minimum–maximum)
25 (6–46)
Iliac rim angle (degrees)
Mean ± standard deviation
23.4 ± 7.5
Median (minimum–maximum)
22 (7–46)
CASTELLVI, n (%)
0
46 (27.5)
1a
9 (5.4)
2a
11 (6.6)
3a
4 (2.4)
1b
55 (32.9)
2b
32 (19.2)
3b
10 (6)
Table 2
Variable: n (%)
Description
(N = 167)
PATGAONKAR frontal approach
Suprailiac
121 (72.5)
Transiliac
46 (27.5)
PATGAONKAR lateral approach
Suprailiac
68 (40.7)
Transiliac
99 (59.3)
CHOI
1
1 (0.6)
2
5 (3)
3
8 (4.8)
4
23 (13.8)
5
37 (22.2)
6
69 (41.3)
7
24 (14.4)
CHOI approach
Suprailiac
37 (22.2)
Suprailiac with foraminoplasty
106 (63.5)
Choi classification, type 7
24 (14.4)
[Table 3 ] shows that the TI approach per the Pantgaonkar classification in lateral radiographs was higher in males than females (p < 0.001). Mean ICH and IRA were higher in patients with an indication for a TI approach per the Pantgaonkar classification (p = 0.001 and p = 0.003, respectively), with an association between the Pantgaonkar profile and the iliac type (p < 0.001). However, after adjusting for gender, age, and iliac type, the mean differences in height and angle were no longer statistically significant (p > 0.05), probably due to the association of the iliac type per Pantgaonkar classification in lateral radiographs.
Table 3
Variable
PATGAONKAR
Lateral approach
p
p *
Suprailiac
Transiliac
Age (years)
0.496**
Mean ± standard deviation
48 ± 15.1
49.8 ± 17.7
Median (minimum–maximum)
47 (18–84)
48 (19–87)
Gender, n (%)
< 0.001
Female
49 (59.8)
33 (40.2)
Male
19 (22.4)
66 (77.6)
Iliac type, n (%)
< 0.001
Android
16 (21.3)
59 (78.7)
Gynecoid
52 (56.5)
40 (43.5)
Iliac crest height (cm)
0.001**
0.130
Mean ± standard deviation
23.7 ± 6.4
27.4 ± 7.9
Median (minimum–maximum)
24 (6–36)
26 (13–46)
Iliac rim angle (degrees)
0.003**
0.134
Mean ± standard deviation
21.4 ± 6.4
24.8 ± 7.9
Median (minimum–maximum)
20.5 (7–42)
24 (11–46)
[Table 4 ] shows that men had a higher frequency of android iliac, while women had a higher frequency of gynecoid pelvis (p < 0.001). Iliac crest height and IRA were higher in patients with an android pelvis (p = 0.002 and p = 0.011, respectively). Nevertheless, after adjusting for the characteristics, only the average iliac height remained statistically higher in the android pelvis (p = 0.039).
Table 4
Variable
Iliac type
p
p *
Android
Gynecoid
Age (years)
0.181**
Mean ± standard deviation
47.1 ± 15.4
50.6 ± 17.6
Median (minimum–maximum)
47 (19–81)
49 (18–87)
Gender, n (%)
< 0.001
Female
1 (1.2)
81 (98.8)
Male
74 (87.1)
11 (12.9)
Iliac crest height (cm)
0.002**
0.039
Mean ± standard deviation
27.9 ± 7.3
24.3 ± 7.3
Median (minimum–maximum)
26 (15–46)
23 (6–46)
Iliac rim angle (degrees)
0.011**
0.078
Mean ± standard deviation
25 ± 7.7
22.1 ± 7.1
Median (minimum–maximum)
24 (11–46)
21 (7–42)
[Table 5 ] shows that the frequency of Choi classification type 7 and TS with foraminoplasty were statistically higher in men than women. Consequently, the frequency of Choi classification type 7 and TS with foraminoplasty was statistically higher in android iliac types (p = 0.001). There was a mean difference in ICH and IRA between the Choi approaches when the values were not adjusted (p = 0.001 and p = 0.004, respectively). However, after adjusting for personal features and iliac type, only ICH showed a statistically significant mean difference (p = 0.042), being statistically higher for TI than SI (p = 0.041) ([Table 6 ]).
Table 5
Variable
CHOI Approach
p
p *
Suprailiac
Suprailiac with foraminoplasty
Choi et al.[7 ]
Age (years)
0.419**
Mean ± standard deviation
46. ± 4.4
49.3 ± 17.8
52 ± 14.6
Median (minimum–maximum)
46 (18–72)
47 (19–87)
52 (23–84)
Gender, n (%)
< 0.001
Female
29 (35.4)
45 (54.9)
8 (9.8)
Male
8 (9.4)
61 (71.8)
16 (18.8)
Iliac type, n (%)
0.001
Android
7 (9.3)
54 (72)
14 (18.7)
Gynecoid
30 (32.6)
52 (56.5)
10 (10.9)
Iliac crest height
0.001**
0.042
Mean ± standard deviation
22.1 ± 6.9
26.5 ± 6.9
29.2 ± 9.1
Median (minimum–maximum)
22 (6–36)
25.5 (13–46)
26.5 (15–46)
Iliac rim angle
0.004**
0.097
Mean ± standard deviation
20.4 ± 7.3
23.7 ± 6.6
26.7 ± 9.9
Median (minimum–maximum)
20 (7–42)
22.5 (11–43)
26 (13–46)
Table 6
Iliac crest height
Comparison
Mean difference
Standard error
p
CI (95%)
Inferior
Superior
Suprailiac x Suprailiac with foraminoplasty
−2.58
1.38
0.192
−5.93
0.77
Suprailiac x Choi classification, type 7
−4.73
1.89
0.041
−9.30
−0.15
Suprailiac with foraminoplasty x Choi classification, type 7
−2.14
1.56
0.513
−5.91
1.63
[Fig. 5 ] shows a high correlation between ICH and IRA for both iliac types (r ≈ 0.9).
Fig. 5 Scatter diagram of the iliac height and the iliac rim angle according to the iliac type and the result of the correlations.
Discussion
The iliac crest is essential in the TF approach at the L5–S1 level, as it prevents disc access in alignment with its axis. This requires a foraminal approach using the SI or TI techniques. Therefore, the morphological study of the pelvis relies on three criteria, that is, ICH, IRA, and pelvic shape.
While Caldwell e Molloy apud Swelson[12 ] categorized the pelvic anatomy into four types—gynecoid, android, platypelloid, and anthropoid—with a focus on assessing the birth canal, our study focused specifically on the android and gynecoid variations to investigate whether the iliac shape could influence the TF approach.
We observed a predominance of the android pelvis in men (87.1%) and gynecoid pelvis in women (98.8%). In addition to a predominance of gynecoid pelvis, females presented a smaller average ICH (24.3 ± 7.3 cm) than males (27.9 ± 7.3 cm), consistent with another study demonstrating that women tend to have a lower iliac crest than men.[13 ] The IRA had no difference in any of the variables studied.
The classification of Patgaonkar et al.[8 ] in lateral radiographs assesses the height of the iliac crest about the L5 vertebra. Both the shape of the pelvis and the gender of the patient demonstrated a significant impact on the choice of the approach. However, after statistical adjustments considering gender and pelvic type, the difference in ICH and IRA becomes insignificant. The fact that female patients had a smaller ICG may have influenced the results before adjustment.
In summary, the Patgaonkar et al.[8 ] classification recommended the SI approach in 63 cases (37.7%) and the TI approach in 104 cases (62.3%). Of the latter, only five cases had an indication for the TI approach based solely on the anteroposterior radiograph. Another 41 cases were suitable for the TI approach in both radiographic views, while 58 were suitable based on the lateral radiographs alone.
The Choi et al.[7 ] classification, which evaluates the height of the iliac crest and the L5 vertebra on lateral radiographs, received an addition, type 7, to encompass cases in which the ICH exceeds the estimates of the original classification of Choi et al.[7 ] As in Patgaonkar et al.[8 ] classification, we observed significant variations related to gender and pelvic shape.
After adjusting for gender and iliac type, we found that the difference in ICH remained significant. The group with indication for an SI approach had a mean ICH of 22.1 ± 6.9 cm, significantly lower than the mean value for the type-7 group (29.2 ± 9.1 cm).
In addition, we evaluated the correlation between ICH and IRA through a scatter plot. We detected a strong correlation (r ≈ 0.9) between the 2 metrics, indicating that an increased ICH corresponds to increased IRA values. Interestingly, the numerical values for ICH and IRA tend to be close to each other.
Finally, 65.8% of the cases analyzed did not present mega-apophyses or showed only type 1a or 1b mega-apophyses patterns according to the Castellvi classification. These patterns have a minimal impact on the TF approach.
Conclusion
The elevated iliac crest is associated with higher grades in the Choi classification. The IRA, although well correlated with ICH, did not show the same statistical difference between the classifications.
