The Lumbar Gap Measurement in Lenke 1-4C Curves

Spine Deformity 6 (2018) 241e249 www.spine-deformity.org The Lumbar Gap Measurement in Lenke 1e4C Curves Hong Zhang, MD*, B. Stephens Richards, MD, Daniel J. Sucato, MD, MS, Chan-Hee Jo, PhD, Dong Tran, MS, Linfeng Wang, MD Texas Scottish Rite Hospital for Children, Department of Orthopaedic Surgery, UT-Southwestern Medical Center at Dallas Received 24 February 2017; revised 27 September 2017; accepted 3 November 2017 Abstract Study Design: Retrospective review. Objectives: To assess whether the lumbar gap (LG) measurement, which is the distance between the center sacral vertical line and the concave edge of the apical vertebra of the lumbar curve, would be a useful tool to predict the need for lumbar curve fusion in the Lenke 1e4C curves. Summary of Background Data: The current treatment guidelines of selective thoracic fusion in the Lenke 1e4C curves are not routinely accepted. Methods: One hundred three adolescent idiopathic scoliosis (AIS) patients had undergone either selective thoracic fusion (STF) or both thoracic and lumbar curves fusion (TLF) for Lenke 1e4C curves. The correlations between the fusion decision making and preoperative LG, coronal balance, thoracic and lumbar Cobb, apical vertebra translation, and rotation were analyzed. The radiographic outcomes and SRS-30 of a minimum 2-year follow-up were reviewed in each group. Results: A total of 51 patients (49.5%) underwent an STF, and 52 patients (50.5%) underwent a TLF. The mean LG was 22.0  8.8 mm in the TLF, which was 2.3 times greater than the STF (9.6  3.9 mm) (p ! .0001). Only 5% of the lumbar curves were fused when the LG was 10 mm or less. Ninety percent of the lumbar curves were fused when the LG was 16 mm or greater, and 100% lumbar curves were fused with an LG of 21 mm or greater. The preoperative coronal imbalance to the left in the TLF was significantly greater than the STF. A mean 47% thoracic correction corresponded to a mean 39% spontaneous correction of the lumbar curve obtained in the SFT, which was significantly different from the TLF (56% and 65%). There were no differences in the SRS-30 scores at 2 years postoperatively between the STF and the TLF. Conclusion: The lumbar curve should not be fused when the LG was 10 mm or less, and very likely should be fused when the LG exceeds 20 mm in the Lenke 1e4C AIS patients. Level of Evidence: Level III Ó 2017 Scoliosis Research Society. All rights reserved. Keywords: Lenke lumbar C modifier; selective thoracic fusion; lumbar gap Introduction The Lenke classification system for adolescent idiopathic scoliosis (AIS) evaluates curve patterns, lumbar apical IRB/Research Advisory Panel Approval: This study was reviewed and approved by the UT Southwestern Institutional Review Board, approval # 022011-127, and the Texas Scottish Rite Hospital for Children Research Advisory Panel, approval # 10-03-670. Author disclosures: HZ (other from Globus Medical, outside the submitted work), BSR (none), DJS (other from Globus, outside the submitted work), CHJ (none), DT (none), LW (none). *Corresponding author. Texas Scottish Rite Hospital for Children, Department of Orthopaedic Surgery, UT-Southwestern Medical Center, 2222 Welborn Street, Dallas, TX 75219, USA. Tel.: (214) 559-7874; fax: (214) 559-7872. E-mail address: [email protected] (H. Zhang). vertebral translation, and sagittal alignment and has been helpful in facilitating surgical decision making [1-3]. In general, for curves in which the lumbar apical vertebral body is touched by the center sacral vertical line (CSVL), termed lumbar modifier A or B, the recommendation is to fuse the main thoracic curve fusion only. The more challenging decision rests with double curves in which the lumbar apical vertebra is totally translated from the CSVL, termed lumbar modifier C. These double curves could also be treated with a selective thoracic fusion (STF) but are considered to be at greater potential for subsequent decompensation. In surgical AIS patients whose deformities include lumbar modifier C curves, the decision to include the lumbar curve in the fusion remains controversial [4-6]. 2212-134X/$ - see front matter Ó 2017 Scoliosis Research Society. All rights reserved. https://doi.org/10.1016/j.jspd.2017.11.001 242 H. Zhang et al. / Spine Deformity 6 (2018) 241e249 Performing an STF for lumbar modifier C double curves requires radiographic and clinical information. Clinical evaluation with the Adam’s forward bending test assesses the severity of the lumbar prominence. Radiographic evaluation of the curve magnitude, apical vertebral translation (AVT), apical vertebral rotation (AVR), and their respective ratio are important [7,8]. A successful STF can often be achieved if their ratios are greater than 1.2. However, this guideline is not routinely adhered to. Previous studies reported that only 6% to 67% of patients with the Lenke 1C curve type underwent STF [6,7,9]. The purpose of this study was to retrospectively review the radiographic results and patient-reported outcomes of the Lenke 1C, main thoracic curve; 2C, double thoracic curve; 3C, double major; and 4C, triple major curves in order to determine which modifier C curves should not be treated by an STF. We hypothesized that the preoperative distance between the CSVL and the concave edge of the apex of the lumbar curve, termed the lumbar gap (LG), would be a predictive indicator of whether or not to include the lumbar curve in the fusion for those curves with a modifier C. Materials and Methods Inclusion criteria and grouping A total of 1,068 AIS cases surgically treated at a single institution between July 2005 and October 2014 were retrospectively reviewed. For this study, inclusion criteria were Lenke 1C, 2C, 3C, and 4C AIS patients surgically treated with a posterior pedicle screweonly constructs with a minimum 2 years’ postoperative follow-up. Those who were treated by all-hook constructs or hybrid constructs (combination hook and screw), and those who had less than 2 years’ follow-up were eliminated. Details of the clinical deformities, radiographic features, and postoperative outcomes was collected on those who qualified for this study. A total of 103 patients were divided into two groups based on the fusion pattern. In the STF group (n 5 51), patients underwent selective thoracic fusion. In the thoracic and lumbar fusion (TLF) group (n 5 52), patients underwent both thoracic and lumbar curve fusion. CT). Measurements were determined on preoperative and follow-up postoperative radiographs. In the standing posterior-anterior (PA) radiograph, the coronal parameters included (Figs. 1 and 2) 1) thoracic (T-Cobb) and lumbar Cobb (L-Cobb) angle; 2) thoracic (T-AVT) and lumbar apical vertebral translation (L-AVT); 3) thoracic (T-AVR) and lumbar (L-AVR) Nash-Moe apical vertebral rotation; 4) coronal balance, which was defined as the distance between a plumb line from the centroid of C7 (C7PL) and the CSVL (C7PL-CSVL), and coronal imbalance, defined as a C7PLCSVL O20 mm; 5) thoracic trunk shift, which was calculated by measuring the distance between the vertical trunk line and the CSVL; 6) shoulder level, which was measured by the clavicle angle, and shoulder imbalance, defined as a clavicle angle O2 ; and 7) coronal position of the lowest instrumented vertebra (LIV), which was the distance between the centroid of LIV and the CSVL (LIVCSVL), was measured 2 months postoperatively and at the final follow-up. The LIV-CSVL change between the 2 months postoperatively and final follow-up was calculated. In the sagittal plane, thoracic kyphosis from T5 to T12, thoracolumbar junction alignment from T10 to L2, and lumbar lordosis from T12 to the sacrum were measured. For the coronal plane radiograph measurements, a positive value indicated either distance translation to the right of the CSVL or left up the tilt angle. A negative value indicated a distance translation to the left of the CSVL or right up the tilt angle. For the sagittal plane measurement, kyphosis was a positive number and lordosis was a negative value. On the preoperative supine bending radiographs, the thoracic and lumbar curves were measured. The thoracic (T-FI) and lumbar flexible index (L-FI) was then calculated: FI 5 standing coronal Cobb angleesupine bending Cobb angle/standing coronal Cobb angle  100%. Lumbar gap measurement The triradiate cartilage, Risser sign, and menarchal status (females) were assessed in each patient. The status of the triradiate cartilage was defined as open, closing (nonfused remnant is incompletely visualized), and closed. The Risser sign was measured in the standard manner from grade 0 to 5. The menarchal history was recorded as premenarchal or postmenarchal (in years). On the standing PA radiograph, the distance between the CSVL and the concave edge of the apex of the lumbar or thoracolumbar curve, termed the lumbar gap (LG), was measured (Fig. 1). To do this, the CSVL was first drawn. The lumbar or thoracolumbar apex, which was the most horizontal and most laterally deviated vertebra or disc from the CSVL, was then identified. The LG was measured from the CSVL to the concave edge of the apex. The LG in which the CSVL touches the apex is defined as zero. The LG was divided into four sections: LG < 10 mm; LG 5 11e15 mm; LG 5 16e20 mm; and LG > 21 mm. The patient distribution was evaluated according to the LG. A possible cut-off value of the LG was analyzed to determine whether or not to include the lumbar curve in the fusion for these patients. Radiographic measurements Patients outcomes All radiographs were evaluated digitally using the Synapse analysis system (Fujifilm Medical Systems, Stamford, The patients completed the Scoliosis Research Society Questionnaire (SRS-30) before and after surgery. Mean Skeletal maturation status H. Zhang et al. / Spine Deformity 6 (2018) 241e249 scores in the domains including pain, function, self-image, mental health, satisfaction, and total score were calculated at preoperation and 2 years postoperation. The mean score in each domain was compared between the two groups. Change in scores between preoperation and 2 years postoperation was calculated. Fig. 1. Lumbar gap (LG): The CSVL is a vertical line that bisects the proximal sacrum. The LG is the distance between the CSVL and the concave edge of the apex of the lumbar or thoracolumbar curve. Coronal balance: The C7PL is the C7 plumb line that is dropped from the middle of the C7 vertebral body and is drawn parallel to the CSVL. The coronal balance (C7PL-CSVL) is the distance between the C7PL and CSVL. Thoracic trunk shift: Through the centroid of the apex of the main thoracic curve, a horizontal line is drawn. The points are marked at the intersection of the horizontal line and the rib cage on the left and right. The vertical trunk reference line (VTRL) is perpendicularly dropped through the midpoint of the line segment. The thoracic trunk shift (TTS) is the distance between the VTRL and the CSVL. 243 Statistical analysis A Mann-Whitney test was used to compare the STF and the TLF at pre- and postsurgery. We also compared pre- and postsurgery variables using a paired t test. For categorical variables, a chi-square test was used to compare between the groups. A percentage of fusion was calculated when the lumbar gap was greater than or equal to a value and it was depicted in a scatter plot. A p value less than .05 was considered significant. Fig. 2. Thoracic apical vertebral translation (T-AVT) is measured from C7PL to the center of the apex of the main thoracic curve. Lumbar apical vertebral translation (L-AVT) is measured from the CSVL to center of the apex of the thoracolumbar or lumbar curve. The clavicle angle is the angle which is subtended between a horizontal reference line and a line which touches the most cephalad aspect of the both the right and left clavicle. 244 H. Zhang et al. / Spine Deformity 6 (2018) 241e249 Results Of 103 AIS patients, 93 were female and 10 male (female-male 5 9.3: 1). The mean age at surgery was 13.8 years, range 11.1e18.5 years. The mean follow-up was Table 1 Skeletal maturation status Girl-boy (ratio) TRC Open Closing Closed Risser Grade 0 Grade 1 and 2 Grade 3e5 Menarchal status (female*) Premenarchal <1.5 years postmenarchal O1.5 years Postmenarchal STF, n (%) (n 5 51) TLF, n (%) (N 5 52) p value 46:5 47:5 O.999 0 1 (2) 50 (98) 0 3 (6) 49 (94) 14 (27) 12 (24) 25 (49) 15 (29) 6 (12) 31 (59) .2634 10 (25) 16 (40) 14 (35) 6 (15) 21 (54) 12 (31) .4031 Fig. 3. Percentage fusion at LG values. .6176 STF, selective thoracic fusion; TLF, thoracic and lumbar fusion; TRC, triradiate cartilage. * Forty and 39 girls had the menstrual history record in the STF and TLF group, respectively. 27.1 months, range 24e70 months. A total of 51 patients (49.5%) underwent the STF and 52 patients (50.5%) underwent the TLF. The distribution of Lenke classification between the STF and TLF was 1C (23 vs. 22 cases), 2C (15 vs. 7 cases), 3C (7 vs. 17 cases), and 4C curves (6 vs. 6 cases). Though we observed more Lenke 2C in STF and more Lenke 3C in TLF, there was no significant difference (p 5 .07). In those with Lenke 3C curve patterns, 21 had lumbar curves and 3 had thoracolumbar curves. The skeletal maturation status is shown in Table 1. There were no Table 2 Preoperative comparisons between the STF and TLF groups STF n LG (mm) L-AVT/LG (ratio) T-Cobb (  ) T-bending (  ) T-FI (%) L-Cobb (  ) L-bending (  ) L-FI (%) T/L-Cobb (ratio) T-AVT (mm) L-AVT (mm) T/L-AVT (ratio) T-Rota (Nash-Moe) L-Rota (Nash-Moe) T/L-Rota (ratio) C7PL-CSVL to the right (mm) C7PL-CSVL to the left (mm) C7PL-CSVL in the center (mm) TTS to the right (mm) TTS to the left (mm) TTS in the center (mm) Sagittal T5eT12 (  ) Sagittal T10eL2 (  ) Sagittal T12eS1 (  ) Left shoulder high (  ) Right shoulder high (  ) Shoulder level (  ) 51 51 51 51 51 51 51 51 51 51 51 51 51 51 51 11 39 1 42 8 1 51 51 51 6 22 23 TLF Mean  SD 9.6 3.1 68.3 38.9 44 52.6 17.3 68 1.3 51.4 27.1 1.9 2.6 2.2 1.4 12.9 14.3  3.9 ± 0.9  14.1  16.9  15 ± 9.5  12.0  20 ± 0.1  13.2  4.2 ± 0.6  0.6  0.7 ± 0.7  8.8  9.7 0 16.7  10.4 5.6  5.2 0 28.1  15.1 0.6  12.5 68.3  14.9 2.3  0.5 3.6  1.6 0.6  0.5 n 52 52 52 52 52 52 52 52 52 52 52 52 52 52 52 8 44 0 24 26 2 52 52 52 4 27 21 p value Mean  SD 22.0 2.0 67.0 42.0 38 61.8 23.9 62 1.1 47.0 40.3 1.2 2.6 2.7 1.0 11.9 22.8  8.8 ± 0.5  10.8  12.7  14 ± 8.8  12.0  17 ± 0.1  11.5  9.2 ± 0.4  0.6  0.5 ± 0.4  10.4  11.8 0 13.0  11.8 11.0  8.0 0 30.0  14.4 2.5  14.4 62.4  13.3 3.0  1.4 3.2  1.1 0.5  0.5 !.0001 !.0001 .5963 .3004 .0229 !.0001 .0057 .1146 !.0001 .0724 !.0001 !.0001 .9254 !.0001 .001 .8354 .0006 N/A .1917 .0856 N/A .252 .2543 .078 .3122 .3516 .5806 AVT, apical vertebral translation; C7PL, plumb line from the centroid of C7; CSVL, center sacral vertical line; FI, flexible index; L, lumbar; LG, lumbar gap; SD, standard deviation; STF, selective thoracic fusion; T, thoracic; T/L, thoracolumbar; TLF, thoracic and lumbar fusion; TTS, thoracic trunk shift. The significantly different p values are in bold. 245 H. Zhang et al. / Spine Deformity 6 (2018) 241e249 Table 3 Postoperative comparisons between the STF and TLF groups STF LG (mm) T-Cobb (  ) T-Cobb correction rate (%) L-Cobb (  ) L-Cobb correction rate (%) T/L-Cobb (ratio) T-AVT (mm) T-AVT correction rate (%) L-AVT (mm) L-AVT correction rate (%) T/L-AVT (ratio) C7PL-CSVL to the right (mm) C7PL-CSVL to the left (mm) C7Pl-CSVL in the center (mm) TTS to the right (mm) TTS to the left (mm) TTS in the center (mm) Sagittal T5eT12 (  ) Sagittal T10eL2 (  ) Sagittal T12eS1 (  ) Left shoulder high (  ) Right shoulder high (  ) Shoulder level (  ) TLF p value n Mean  SD n Mean  SD 51 51 51 51 51 51 51 51 51 51 51 3 45 3 11 37 3 51 51 51 14 9 28 5.4  6.1 36.0 ± 11.0 47  14 32.0 ± 9.0 39  14 1.2  0.3 28.3 ± 13.5 43  26 23.4 ± 9.8 16  41 1.5  1.8 7.6  5.7 L17.4 ± 10.4 0 8.0  5.9 13.2  5.7 0 24.5  8.1 2.8  13.2 60.8  24.0 3.4  1.6 2.3  0.5 0.8  0.4 52 52 52 52 52 52 52 52 52 52 52 9 41 2 14 32 6 52 52 52 20 6 25 2.0  4.1 29.6 ± 11.8 56  14 22.5 ± 11.5 65  15 1.9  2.8 21.4 ± 11.1 53  24 12.8 ± 13.6 69  31 2.2  3.0 6.1  6.4 L12.7 ± 8.8 0 6.9  4.2 10.4  8.2 0 22.0  7.8 10.1  6.4 63.9  12.3 2.9  1.2 3.5  1.9 0.7  0.5 .0013 .0054 !.0001 !.0001 !.0001 .0616 .0063 .0387 !.0001 !.0001 .1533 .7229 .0268 N/A .5841 .1087 N/A .111 .0006 .4143 .1466 .0939 .8092 AVT, apical vertebral translation; C7PL, plumb line from the centroid of C7; CSVL, center sacral vertical line; L, lumbar; LG, lumbar gap; SD, standard deviation; STF, selective thoracic fusion; T, thoracic; T/L, thoracolumbar; TLF, thoracic and lumbar fusion; TTS, thoracic trunk shift. The significantly different p values are in bold. The ratios of T/L-Cobb and T/L-Rota in the TLF group were less than 1.2 and were significantly less than those of the STF group. The C7PL-CSVL shift to the left in the TLF group was greater than that in the STF group (p 5 .0006). More patients in the TLF group (40%) had coronal imbalance O20 mm to the left than was found in the STF group (20%) (p 5 .0215). Twelve of 52 patients (23%) in the TLF had a thoracic trunk shift O10 mm to the left, which was significantly different from the STF (1 of 51 patients, 2%). significant differences in the triradiate cartilage, Risser sign, and menarchal status between the two groups. Preoperative evaluation The preoperative measurements comparing the STF group with the TLF group are shown in Table 2. There were several significant differences found between those who underwent STF compared with those who underwent TLF. The mean LG in the TLF group was 2.3 times greater than that in the STF group (p ! .0001). The mean L-Cobb in the TLF was 15% greater than that in the STF (p ! .0001). The mean L-AVT in the TLF group was 33% greater than that in the STF group (p ! .0001). The mean L-Rota in the TLF group was significantly greater than that in the STF group. Postoperative evaluation The incidence of performing a TLF was 5% (2 of 38 cases) in the LG <10 mm, 52% (12 of 23 cases) in the Table 4 Comparison of the preoperative and 2 years postoperative SRS-30 score and its domains between STF and TLF STF TLF Preop Pain Function Self-image Mental health Satisfaction Total score 3.9 4.0 3.3 3.9 3.8 3.8       0.7 0.5 0.6 0.7 0.8 0.4 2 years postop p value Preop 4.1 4.2 4.1 4.0 4.5 4.1 .5034 .3508 !.0001 .4288 .0059 .0944 4.1 4.0 3.4 4.1 3.8 3.9       0.7 0.6 0.6 0.9 0.5 0.5       Group comparison 0.7 0.6 0.5 0.6 0.9 0.4 2 years Postop p value Preop. p value Postop. p value 4.2 4.3 4.2 4.3 4.6 4.2 .8368 .0897 !.0001 .174 .0002 .0133 .1661 .9972 .6451 .262 .7538 .2579 .8248 .4841 .3269 .0987 .8771 .245       0.9 0.4 0.5 0.6 0.6 0.4 Postop., postoperative; Preop., preoperation; SRS-30, Scoliosis Research Society, 30-item; STF, selective thoracic fusion; TLF, thoracic and lumbar fusion. 246 H. Zhang et al. / Spine Deformity 6 (2018) 241e249 LG 5 11e15 mm, 77% (13 of 17 cases) in the LG 5 16e20 mm, and 100% (25 of 25 cases) in the LG >21 mm (p ! .0001). The possible cutoff value analysis showed that the 90% lumbar curves were fused when the LG was greater than 16 mm, and 100% were fused with an LG >21 mm (Fig. 3). The postoperative measurements comparing the STF group and the TLF group are shown in Table 3. Significant differences were found between the groups in the mean T-Cobb correction rate and L-Cobb correction rate, the mean T-AVT correction rate and L-AVT correction rate, and the mean postoperative coronal balance (C7PL-CSVL). The self-image and satisfaction scores in both groups had significantly improved at 2 years postoperation compared to preoperation (Table 4). The total score significantly improved in the TLF but not in the STF group. Discussion In 1983, King et al. recommended the STF for the King-Moe classification Type II curve, which is an S-shaped double curve pattern that comprised the right thoracic and left lumbar curve that cross the midline and the thoracic curve is equal to or larger and more rigid than the lumbar curve [10]. In 1992, Lenke et al. reported that the King-Moe type II curve was not sufficient to recommend the STF. The improper identification of true double major curves and improper selection of the distal fusion level was suspected to be the primary cause of most of the lumbar curve decompensations [11]. In 2001, Lenke et al. presented a new classification system for AIS based on the Cobb bending magnitude and taking into account both the coronal and sagittal planes [1]. Performing an STF would be recommended based on several elements of the Lenke classification system, such as Cobb magnitude, AVT, and AVR and their respective ratio 1.2, and contain curves with C modifiers. However, these treatment guidelines are not routinely accepted. In patients with Lenke 1C curves, for instance, only the thoracic curve should undergo fusion by the classification because the lumbar curve is not a true structural curve (lumbar bending !25 ) [8]. Yet many surgeons still prefer to fuse both curves for some 1C curves. Newton et al reported that five AIS centers in the United States varied between 6% and 67% for selection thoracic fusions for Lenke 1C curves [6,7,9]. Our study showed that the STF was not performed in 43% of patients with the Lenke 1C and 2C curves in which the lumbar curve was a nonstructural curve. In addition, our data showed that the STF was performed in 36% of patients with Lenke 3C and 4C curves in spite of the fact that the lumbar curves were the structural curve. Therefore, our study demonstrated that the STF does not fully correlate with the Lenke classification in 1e4C curves. Our study demonstrated that the LG highly correlated with the decision on whether or not the lumbar curve was fused in the Lenke 1e4C curves. When the LG was <10 mm, the lumbar curve was almost never fused (Fig. 4). If the LG was >16 mm, 90% patients’ lumbar curves were fused. If the LG was >21 mm, the lumbar curve was fused every time (Figs. 5 and 6). This study indicated that there was a higher likelihood of fusing both thoracic and lumbar curves with increasing amounts of left coronal imbalance and left thoracic trunk shift. If the coronal imbalance to the left exceeds 20 mm and the thoracic trunk shift to the left exceeds 10 mm, fusion of both curves should be considered. Fig. 4. Representative example of a selective thoracic fusion in a Lenke 1C curve (14-year-2-month-old girl) before surgery (AeC) and 2 years postoperation (D, E). The Lenke 1C curve case represented in this figure demonstrated a preoperative lumbar Cobb angle of 57 and supine bending of 20 . The preoperative lumbar gap (LG) was 9.2 mm. The thoracic trunk shift (TTS) was 17.9 mm to the right. The coronal balance was perfect. At 2 years postoperation, the thoracic curve was corrected to 32 (57%) and the lumbar curve obtained the 53% spontaneous correction. The postoperative radiographs demonstrated well-maintained coronal balance and sagittal alignment. H. Zhang et al. / Spine Deformity 6 (2018) 241e249 247 Fig. 5. Both thoracic and lumbar fusion (TLF) in a Lenke 1C curve (16-year-3-month-old girl). The preoperative thoracolumbar Cobb angle was 60 , which reduced to 19 on the supine bend radiograph. The preoperative lumbar gap (LG) was 30 mm. Coronal imbalance was 27.3 mm to the left. The thoracic trunk shift (TTS) was 10.1 mm to the left (AeC). At 2 years postoperation, the thoracic curve was corrected to 11 (82%), and the lumbar curve obtained the 83% correction. The postoperative radiographs demonstrated well-maintained coronal balance and sagittal alignment (D, E). The flexibility of the lumbar curve on bending films has been suggested as a determinant for lumbar fusion in the lumbar modifier C curves [8]. However, our study showed that no correlation was found between the lumbar fusion and the lumbar flexibility. Our results suggested that the lumbar curve’s flexibility was not a reliable predictor for the lumbar fusion in the Lenke 1e4C curves. Several factors have been considered important in making the decision to perform an STF or fuse both curves in Lenke 1e4C curves. The Cobb magnitude, AVT, and AVR of the thoracic and lumbar curve and their ratio greater than 1.2 has been suggested as important. Our results confirmed there was a correlation between the ratios of Cobb, AVT, and AVR and the fusion decision making of the lumbar curve. But the O1.2 ratio of the T/L-AVT may be an inadequate predictor for the STF because more than 50% patients in the TLF had a O1.2 T/L-AVT in this study. Although the LG is similar in concept to the L-AVT, it has certain advantages. The LG is a qualitative parameter for the lumbar curve modifiers in that it applies only to the lumbar C modifier, and aids in deciding if the lumbar curve needs to be included in the fusion. The L-AVT does not distinguish between the lumbar modifiers. Two curves with different lumbar modifiers may have the same L-AVT Fig. 6. Both thoracic and lumbar fusion (TLF) in a Lenke 3C curve (14-year-6-month-old girl). The preoperative thoracolumbar Cobb angle was 76 , which reduced to 41 on the supine bend radiograph. The preoperative lumbar gap (LG) was 34.2 mm. The coronal imbalance to the left measured 44.8 mm. The thoracic trunk shift (TTS) was 5.9 mm to the left (AeC). At 2 years postoperative, the thoracic curve was corrected to 15 (82%) and the thoracolumbar curve obtained the 86% correction. The postoperative radiographs demonstrated well-maintained coronal balance and sagittal alignment (D, E). 248 H. Zhang et al. / Spine Deformity 6 (2018) 241e249 Fig. 7. (A) A 13-year-11-month-old girl with AIS. The lumbar apical vertebra (L2) is totally translated from the CSVL. The lumbar curve is a C modifier. The lumbar apical vertebral rotation is grade 3 (Nash-Moe). The LG measurement is 6.5 mm and L-AVT is 21.8 mm. (B) showing a 14-year-9-month-old girl with AIS. The CSVL touches the pedicle area of the lumbar apical vertebra (L3). The lumbar curve is a B modifier that has no LG. The lumbar apical vertebral rotation is grade 1 (Nash-Moe), which was less than the one in the patient A. L-AVT measurement is 21.7 mm, which is almost the same as the one in patient A. The patients in panels A and B have a same L-AVT but a different lumbar modifier. (Fig. 7). In addition, the L-AVT measurement is influenced by the lumbar vertebral rotation, as the width of the vertebra on the PA radiograph varies with changes in vertebral rotation. Our study indicated that the mean thoracic correction rate was 47% and the lumbar curve had a mean 39% spontaneous correction in the STF which, not surprisingly, was lower than the correction achieved with TLF. However, there was no difference in the postoperative T/L-Cobb ratio and the T/L-AVT ratio between the two groups. These findings mean that the STF can provide a similar balanced correction of the thoracic and lumbar curves with the TLF. There was also a similar correction in the sagittal profiles between the two groups. We did note a trend toward greater postoperative coronal imbalance to the left in the STF. Our results found that no postoperative differences were seen in the SRS-30 scores between the STF and the TLF. Within each group, the self-image and satisfaction scores had improved following the surgery. With this information, STF should be considered as the primary choice whenever possible (Fig. 8). Preserving lumbar spine motion with an STF portends a more favorable long-term prognosis. Limitations of this study include the following: 1. This is a relatively short-term follow-up retrospective study. The patients’ outcomes (SRS-30) were collected at 2 years postoperation. Longer-term follow-up, for example, 20 or 30 years, will be needed to further validate the conclusions, especially for the patients with double major curves who underwent STF. 2. It must be remembered that this is a retrospective study comparing the STF and TLF groups on LG measurement and the findings. There are discrepancies in the numbers in Lenke 2C and 3C groups between the STF and TLF, which makes comparison H. Zhang et al. / Spine Deformity 6 (2018) 241e249 249 Fig. 8. Selective thoracic fusion (STF) in a Lenke 3C curve (12-year-old girl, Risser 0, 8 months postmenarchal). The preoperative lumbar Cobb angle was 67 , which reduced to 47 on the supine traction radiograph. The preoperative lumbar gap (LG) was 13.2 mm. The coronal balance was perfect. The thoracic trunk shift (TTS) was 23.9 mm to the right (AeC). At 2 years postoperation, the thoracic curve was corrected to 39 (58%) and the lumbar curve obtained the 57% spontaneous correction. The postoperative radiographs demonstrated well-maintained coronal balance and sagittal alignment (D, E). difficult. This study reports the results found, but cannot recommend different treatments without a matching study, including curve magnitudes and flexibilities. 3. The present study focused on radiographic analysis for decision making on fusion level selection. We recognize that choosing fusion levels in these double curves is not only related to radiographic evaluation but also dependent on the clinical examination and the patient’s input. In conclusion, preoperative assessment of the LG may be helpful in making the decision whether or not to include a lumbar fusion in the Lenke 1e4C AIS patients. The lumbar curve should not be fused when the LG is <10 mm, and very likely should be fused when the LG exceeds 20 mm. When the LG is between 10 and 20 mm, other factors, such as coronal balance to the left O20 mm, and thoracic trunk shift to the left O10 mm should be included in the decision making regarding the lumbar fusion. Key points Lumbar gap (LG) is the distance between the center sacral vertical line and the concave edge of the apical vertebra of the lumbar curve in the Lumbar C modifier. There is a correlation between the fusion decision making and the LG in the Lenke 1e4C curves. The lumbar curve should not be fused when the LG is 10 mm or less, and very likely should be fused when the LG exceeds 20 mm in the Lenke 1e4C AIS patients. References [1] Lenke LG, Betz RR, Harms J, et al. Adolescent idiopathic scoliosis: a new classification to determine extent of spinal arthrodesis. J Bone Joint Surg Am 2001;83-A:1169e81. [2] Lenke LG, Betz RR, Clements D, et al. Curve prevalence of a new classification of operative adolescent idiopathic scoliosis does classification correlate with treatment. Spine 2002;27:604e11. [3] Lenke LG, Edwards 2nd CC, Bridwell KH. The Lenke classification of adolescent idiopathic scoliosis: how it organizes curve patterns as a template to perform selective fusion of the spine. Spine 2003;28: S199e207. [4] Edwards 2nd CC, Lenke LG, Peelle M, et al. Selective thoracic fusion for adolescent idiopathic scoliosis with C modifier lumbar curves: 2- or 16-year radiographic and clinical results. Spine 2004;29:536e46. [5] Ritzman TF, Upasani VV, Bastrom TP, et al. Comparison of compensatory curve spontaneous derotation after selective thoracic or lumbar fusions in adolescent idiopathic scoliosis. Spine 2008;33:2643e7. [6] Fischer CR, Kim Y. Selective fusion for adolescent idiopathic scoliosis: a review of current operative strategy. Eur Spine J 2011;20: 1048e57. [7] Crawford III CH, Lenke LG, Sucato DJ, et al. Selective thoracic fusion in Lenke 1C curves. Spine 2013;38:1380e5. [8] Schulz J, Asghar J, Bastrom T, et al. Optimal radiographical criteria after selective thoracic fusion for patients with adolescent idiopathic scoliosis with a C lumbar modifier. Spine 2014;39:E1368e73. [9] Newton PO, Faro FD, Lenke LG, et al. Factors involved in the decision to perform a selective versus nonselective fusion of Lenke 1B and 1C curves in adolescent idiopathic scoliosis. Spine 2003;28: S217e23. [10] King HA, Moe JH, Bradford DS, et al. The selection of fusion levels in thoracic idiopathic scoliosis. J Bone Joint Surg Am 1983;65-A: 1302e13. [11] Lenke LG, Bridwell KH, Baldus C, et al. Preventing decompensation in King Type II curves treated with Cotrel-Dubousset instrumentation. Strict guideline for selective thoracic fusion. Spine 1992;17: S274e81.