Comparison of anterior decompression and fusion versus laminoplasty in the treatment of multilevel cervical ossification of the posterior longitudinal ligament: a systematic review and meta-analysis.

PURPOSE: A meta-analysis was conducted to evaluate the clinical outcomes, complications, reoperation rates, and late neurological deterioration between anterior decompression and fusion (ADF) and laminoplasty (LAMP) in the treatment of multilevel cervical ossification of the posterior longitudinal ligament (OPLL).
METHODS: All related studies published up to August 2015 were acquired by searching PubMed and EMBASE. Exclusion criteria were case reports, revision surgeries, combined anterior and posterior surgeries, the other posterior approaches including laminectomy or laminectomy and instrumented fusion, non-English studies, and studies with quality assessment scores of <7. The main end points including Japanese Orthopedic Association (JOA) score, recovery rate of JOA, cervical lordosis, complication rate, reoperation rate, and late neurological deterioration were analyzed. All available data was analyzed using RevMan 5.2.0 and Stata 12.0.
RESULTS: A total of seven studies were included in the meta-analysis. The mean surgical level of ADF was 3.1, and the mean preoperative occupation ratios of ADF and LAMP group were 55.9% and 51.9%, respectively. No statistical difference was observed with regard to preoperative occupation ratio and preoperative JOA score. Although LAMP group had a higher preoperative cervical lordosis than ADF group (P<0.05, weighted mean difference [WMD] =-5.73, 95% confidence interval [CI] =-9.67--1.80), significantly decreased cervical lordosis was observed in LAMP group after operation. ADF group had higher postoperative JOA score (P<0.05, WMD =2.18, 95% CI =0.98-3.38) and neurological recovery rate (P<0.05, WMD =27.22, 95% CI =15.20-39.23). Furthermore, ADF group had a lower late neurological deterioration rate than the LAMP group (P<0.05, risk difference =0.16, 95% CI =0.04-0.73). The complication rates of both groups had no statistical difference. However, LAMP group had a significantly lower reoperation rate than ADF group. The reoperation rate of ADF group (20.5%) was almost six times that of LAMP group (3.5%).
CONCLUSION: Our meta-analysis suggested that ADF was associated with better postoperative neurological function, neurological recovery rate, and less late neurological deterioration than LAMP in the treatment of multilevel cervical OPLL with a high mean occupation ratio. LAMP was associated with a decreased postoperative cervical lordosis, which might be a cause of late neurological deterioration. The complication rates of both groups showed no statistical difference. However, the reoperation rate was significantly higher in ADF group compared with LAMP group. Benefits and risks should be balanced when ADF or LAMP is selected.

Introduction

Ossification of the posterior longitudinal ligament (OPLL) is an important cause of cervical myelopathy. The incidence of OPLL ranges from 1.9% to 4.3% in East Asian countries and from 0.01% to 1.7% in Caucasian populations.1,2 Although many clinical features of cervical OPLL are similar to those of cervical myelopathy caused by cervical disc herniation, the former still has several unique characteristics. Conservative treatment is usually ineffective for moderate-to-severe myelopathy caused by OPLL; instead, surgical treatment is the first option in these cases.

There are two representative surgical approaches: anterior decompression and fusion (ADF) and laminoplasty (LAMP). However, some controversies still remain on surgical selection. Removal of OPLL from anterior approach seems to be radical to decompress the spinal cord, and fusion can establish a cervical stability that is conducive to relieve pressure on the levels of compressed cervical cord.3 Meanwhile, complications and technical problems, including dural tear, graft extrusion, insufficient decompression, and so on, are still great concerns, especially when multiple levels and high occupation ratios are involved.4,5 ADF remains a significant surgical technical challenge.

The indirect decompression via LAMP and cervical lordosis alignment allows the spinal cord to float away from ventral compression. However, if posterior shift of the cord is insufficient, ventral constriction of the cord may persist, leading to diminished recovery of neurological function. Admittedly, the surgical technique is less difficult for LAMP than ADF.

At present, no standards or guidelines exist for the treatment of OPLL. We performed this meta-analysis to evaluate the clinical outcomes, complications, reoperation rates, and late neurological deterioration between ADF and LAMP in the treatment of cervical OPLL with the aim of trying to find the evidences for how to balance the benefits and risks of the aforementioned surgical approaches.

Methods

We developed a protocol prior to this systematic review, which was registered in PROSPERO. The registration number is CRD42015025032.

Search strategy

The primary sources of the studies reviewed in this meta-analysis were PubMed and EMBASE. The search included literature exclusively in English and published up to August 2, 2015. The following terms were used in our search: anterior AND (ossification of the posterior longitudinal ligament OR ossified posterior longitudinal ligament OR calcification of the posterior longitudinal ligament). Reference lists of all included studies were scanned to identify potentially relevant studies. Two reviewers independently screened the titles and abstracts of studies identified from the search. Full-text copies of all potentially relevant studies were obtained.

Inclusion and exclusion criteria

Studies were included if they met the following criteria: 1) study design: prospective or retrospective comparative study; 2) patients with cervical myelopathy due to OPLL, excluding patients with tumors, trauma, infection, previous surgeries, revision surgeries, combined anterior and posterior surgeries, and other posterior approaches including laminectomy or laminectomy and instrumented fusion; 3) purpose of the studies: to compare clinical outcome differences between ADF and LAMP; 4) outcome measurements: including Japanese Orthopedic Association (JOA) score, neurological recovery rate, cervical lordosis, complications, reoperation rate, and late neurological deterioration; and 5) published in English. Studies that did not meet these criteria were excluded.

Data extraction

The following information was extracted from each study: 1) study ID, 2) study design, 3) study location, 4) number of cases, 5) length of follow-up, 6) number of surgical levels, 7) preoperative occupation ratios, 8) space available for spinal cord, 9) operation time, 10) blood loss, 11) preoperative and postoperative JOA scores, 12) cervical lordosis, 13) neurological recovery rate, 14) complications, 15) reoperations, 16) late neurological deterioration, and 17) patient’s age.

Data analysis

Statistical analysis was conducted using the Review Manager software (RevMan Version 5.2. The Nordic Cochrane Center, The Cochrane Collaboration, Copenhagen, Denmark). Heterogeneity was tested using the chi-square test and quantified by calculating the I2 statistic, in which P<0.05 and I2>50% were considered statistically significant. For the pooled effects, weighted mean difference (WMD) was calculated for continuous variables according to the consistency of measurement units, and odds ratio (OR) was calculated for dichotomous variables. Continuous variables are presented as mean differences and 95% confidence intervals (CIs), whereas dichotomous variables are presented as ORs and 95% CI. Random-effects or fixed-effects models were used depending on the heterogeneity of the studies included. Publication bias was tested using a funnel plot.

Results

Characteristics of studies

A total of 951 papers were first identified by screening the titles and abstracts. Of these, 935 papers were excluded because they were duplicates, irrelevant studies, case reports, revision surgeries, combined anterior and posterior surgeries, non-English studies, and reviews. The remaining 16 papers underwent a detailed and comprehensive evaluation (Figure 1). Papers with quality assessment scores of <7 and all other posterior approaches including laminectomy or laminectomy and instrumented fusion were excluded. The remaining seven studies were finally included in this meta-analysis.6–12 Details of these studies are listed in Table 1.

Figure 1

Flowchart of study selection.

Table 1

Characteristics and quality assessment of included studies

Source	Design	Country	Number of cases	Surgical levelsa	Occupation ratio (%)a	Space of cord (mm)a	Follow-up timea	Operation time (min)a	Blood loss (mL)a	Agea	Newcastle–Ottawa scale
Fujimori et al6	Nonrandomized retrospective comparative	Japan	A: 12L: 15	A: 3.3±0.9L: 5.4±1.2	A: 67.5±6.4L: 66.0±6.5	A: 4.0±1.0L: 5.4±1.4	A (year): 9.9±4.1L (year): 10.2±5.7	A: 314±89.9L: 128±60	A: 600±459L: 240±153	A: 55.6±7.8L: 58.7±9.1	8
Chen et al7	Nonrandomized retrospective comparative	People’s Republic of China	A: 22L: 25	A: 2.8bL: N/A	A: 55.4±8.2L: 54.3±4.6	N/A	A (year): >4L (year): >4	N/A	N/A	A: 57.2bL: 54.2b	7
Sakai et al8	Nonrandomized prospective comparative	Japan	A: 20L: 22	A: 3.1bL: 4.5b	A: 43.4±16.6L: 46.9±16.1	N/A	A (year): 5bL (year): 5b	A: 300.3±78.6L: 183.2±41.1	A: 292.8±192.8L: 289.6±215.8	A: 59.5±9.3L: 58.4±9.6	8
Lee et al9	Nonrandomized retrospective comparative	South Korea	A: 20L: 27	A: 3.3±0.3L: 5.5±0.6	N/A	A: 5.4±1.3L: 5.1±1.7	A (month): 19.8±12.3L (month): 29.1±15.3	N/A	N/A	A: 56.4±8.5L: 54.7±10.5	7
Iwasaki et al10	Nonrandomized retrospective comparative	Japan	A: 27L: 66	A: 3bL: N/A	A: 56.6bL: 44.4b	A: 5.6bL: 7.6b	A (year): 6bL (year): 10.2b	A: 302bL: 177b	A: 513bL: 464b	A: 57bL: 58b	8
Masaki et al11	Nonrandomized retrospective comparative	Japan	A: 19L: 40	A: 2.9±0.9L: 4.6±0.5	A: 56.0±8.0L: 55.9±14.3	N/A	A (year): >1L (year): >1	N/A	N/A	A: 51.8±6.6L: 62.6±10.3	8
Tani et al12	Nonrandomized retrospective comparative	Japan	A: 14L: 12	A: 3.3±0.6L: 5.6±0.8	A: 62.9±7.6L: 66.7±8.7	N/A	A (month): 49.1±34.5L (month): 50.3±42.7	N/A	N/A	A: 61.5±11.4L: 65.8±6.1	8

Notes:

Data are represented as mean ± standard deviation unless otherwise indicated.

Only mean values were reported.

Abbreviations: N/A, not applicable; A, anterior decompression and fusion; L, laminoplasty; min, minutes.

Quality assessment

Two investigators evaluated each study and extracted data independently; any disagreements were resolved by discussion. The baseline characteristics of the participants in these studies were similar. Newcastle–Ottawa quality assessment scale was used to assess the quality of the included studies. Among these studies, five of them scored 8 points and two scored 7 points. These scores indicate that the studies included in this meta-analysis were of high quality (Table 1).

Surgical level and surgical approach selection

All included papers reported the surgical levels of ADF. The mean surgical level of ADF was 3.1. The surgical levels of LAMP are shown in Table 1. Only one study selected surgical approach according to different given time of surgery (ADF group in 1997, 1999, 2001, 2003, and 2004 and LAMP group in 1996, 1998, 2000, and 2002),8 while the other six studies had no clear or definitive selection criterion for each surgical approach.

Preoperative occupation ratios

The preoperative occupation ratios were reported in six studies. Approximately 294 cases were involved, including 114 cases of ADF and 180 cases of LAMP. The mean preoperative occupation ratios were 55.9% in ADF group and 51.9% in LAMP group. There was no significant difference in the preoperative occupation ratio between ADF group and LAMP group (P=0.474).

Preoperative and postoperative JOA scores

Preoperative and postoperative JOA scores (final follow-up) were analyzed in six studies. Standard deviation of one study was not reported; thus, the statistical data for this one were unavailable. Five studies and 201 cases were involved, including 87 cases of ADF and 114 cases of LAMP. There was no significant difference in the preoperative JOA score between ADF group and LAMP group (P>0.05, WMD =0.54 [−0.05, 1.12]; Figure 2), and the chi-square test indicated no statistical evidence of heterogeneity (I2=0%, P=0.07). The ADF group had a significant higher postoperative JOA score than the LAMP group (P<0.05, WMD =2.18 [0.98, 3.38]; Figure 2), and moderate heterogeneity existed between these studies (I2=76%, P=0.0004).

Figure 2

Forest plot illustrating the comparison in preoperative JOA score (fixed-effects model, 2.1), the postoperative JOA score (random-effects model, 2.2), and recovery rate of JOA (random-effects model, 2.3) between ADF and LAMP groups.

Abbreviations: CI, confidence interval; df, degrees of freedom; IV, independent variable; SD, standard deviation; JOA, Japanese Orthopedic Association; ADF, anterior decompression and fusion; LAMP, laminoplasty.

Neurological recovery rate

The recovery rates of JOA at final follow-up were analyzed in six studies. Standard deviation of one study was not reported, and the statistical data were unavailable and thus excluded. Five studies and 201 cases were involved, including 87 cases of ADF and 114 cases of LAMP. The mean recovery rate of ADF and LAMP groups was 63.9% and 40.0%, respectively. The ADF group had a significantly higher postoperative neurological recovery rate than LAMP group (P<0.05, WMD =27.22 [15.20, 39.23]; Figure 2). Moderate heterogeneity existed between these studies (I2=68%, P<0.00001).

Preoperative and postoperative cervical lordosis

Preoperative and postoperative cervical lordosis (final follow-up) were analyzed in three studies. A total of 116 cases were involved, including 52 cases of ADF and 64 cases of LAMP. The LAMP group had a significantly higher preoperative cervical lordosis than the ADF group (P<0.05, WMD =−5.73 [−9.67, −1.80]; Figure 3), and mild heterogeneity existed between these studies (I2=10%, P=0.004). There was no significant difference in the postoperative cervical lordosis between ADF group and LAMP group (P>0.05, WMD =2.05 [−6.23, 10.34]; Figure 3), and moderate heterogeneity existed between these studies (I2=81%, P=0.63). The postoperative cervical lordosis was significantly lower than preoperative cervical lordosis in LAMP group (P<0.05, WMD =−4.05 [−7.80, −0.30]). There was no significant difference between preoperative and postoperative cervical lordosis in ADF group (P>0.05, WMD =3.09 [−0.67, 6.84]).

Figure 3

Forest plot illustrating the comparison in preoperative cervical lordosis (fixed-effects model, 3.1) and postoperative cervical lordosis (random-effects model, 3.2) between ADF and LAMP groups.

Abbreviations: CI, confidence interval; df, degrees of freedom; IV, independent variable; SD, standard deviation; ADF, anterior decompression and fusion; LAMP, laminoplasty.

Late neurological deterioration

Late neurological deterioration of LAMP and ADF was reported in four and five studies, respectively. Four studies and 174 cases were involved, including 75 cases of ADF and 99 cases of LAMP. The LAMP group had a significantly higher incidence of late neurological deterioration than ADF group (P>0.05, OR =0.16 [0.04, 0.73]; Figure 4). The chi-square test indicated no statistical evidence of heterogeneity (I2=0%, P=0.02).

Figure 4

Forest plot (fixed-effects model) illustrating the comparison in late neurological deterioration (4.1), complication (4.2), and reoperation (4.3) between ADF and LAMP groups.

Abbreviations: CI, confidence interval; df, degrees of freedom; M–H, Mantel–Haenszel; SD, standard deviation; ADF, anterior decompression and fusion; LAMP, laminoplasty.

Complications

Complications of LAMP and ADF were reported in four and five studies, respectively. Four studies and 163 cases were involved, including 74 cases of ADF and 89 cases of LAMP. There was no significant difference in complication rate between ADF group and LAMP group (P>0.05, OR =1.46 [0.69, 3.10]; Figure 4). Mild heterogeneity existed between these studies (I2=28%, P=0.32).

Reoperations

Reoperations were reported in four studies. A total of 188 cases were involved, including 73 cases of ADF and 115 cases of LAMP. The mean reoperation rates of ADF and LAMP group were 20.5% and 3.5%, respectively. The ADF group had a significantly higher reoperation rate than LAMP group (P<0.05, OR =5.99 [2.02, 17.83]; Figure 4). Mild heterogeneity existed between these studies (I2=19%, P=0.001).

Publication bias

The Stata 12.0 software (StataCorp LP, College Station, TX, USA) was used to examine the publication bias of the main results. All funnel plots were largely symmetrical including preoperative and postoperative JOA and cervical lordosis, neurological recovery rate, and complication rate (Figure 5). These results indicated that publication bias did not play a vital role in the observed effects, and the conclusions were reliable.

Figure 5

Funnel plot analysis of studies on preoperative JOA score (A), postoperative JOA score (B), recovery rate of JOA (C), preoperative cervical lordosis (D), postoperative cervical lordosis (E), late neurological deterioration (F), complication (G), and reoperation (H) that shows publication bias.

Abbreviations: JOA, Japanese Orthopedic Association; MD, mean difference; OR, odds ratio; SE, standard error; RD, risk difference.

Discussion

The ideal surgical treatment option for multilevel cervical OPLL remains controversial and presents a significant surgical challenge. Previously, a systematic review that included only three papers had been published regarding decision making in the treatment of OPLL, and no definitive conclusion was reached.13 Five meta-analysis papers about surgery approaches of cervical myelopathy have also been published,14–18 but two of them excluded the studies about OPLL,14,15 the other papers had not analyzed the OPLL separately, and the number of included studies about OPLL was equal to four or less.16–18 Moreover, the five meta-analysis papers published had conflicting results, and no consensus on neurological recovery rate had been reached. Therefore, we performed the meta-analysis to evaluate the clinical outcomes, complications, reoperations, and late neurological deterioration between ADF and LAMP in the treatment of cervical OPLL.

In our meta-analysis, seven included papers were considered to be of sufficient methodological quality. The mean surgical level of ADF was 3.1, and the mean preoperative occupation ratios of ADF and LAMP group were 55.9% and 51.9%, respectively. No statistical difference was observed in preoperative occupation ratios and preoperative JOA scores. It indicated that the baseline neurological function characteristics of these two groups were similar. However, ADF group had higher postoperative JOA score and recovery rate. The mean recovery rates of ADF group and LAMP group were 63.9% and 40.0%, respectively. It indicated that the direct decompression of ADF was associated with better neurological recovery and function in the treatment of cervical OPLL with a high mean occupation ratio.

LAMP relies heavily on the decompression effect for indirect decompression through a posterior shift of the spinal cord, and cervical lordosis plays an important role in preventing the static compression. Yamazaki et al19 reported that cervical lordosis of <10° or OPLL thickness of >7 mm were risk factors for spinal cord contact with OPLL. Fujimori et al6 concluded that preoperative cervical lordosis of 20° or more might be necessary if LAMP was applied to successfully treat OPLL with an occupation ratio of 60% or higher. The K-line was also advocated to evaluate both cervical alignment and the size of OPLL. A negative K-line slope means that the OPLL touches the spinal cord and has a significantly lower recovery rate than the positive K-line group.20 In our meta-analysis, the LAMP group had a higher preoperative cervical lordosis than ADF group.

Furthermore, postoperative kyphotic change after LAMP has also been detected. Sakai et al8 reported that it was observed in 50% of the LAMP at 5-year follow-up.8 Biomechanical analysis of cervical OPLL indicated that stress distribution increased with the progression of kyphosis after posterior decompression, which was likely associated with the late neurological deterioration.21 In our meta-analysis, although LAMP group had a higher preoperative cervical lordosis than ADF group, cervical lordosis of LAMP group had significantly decreased after operation. The preoperative and postoperative cervical lordosis of ADF group and the postoperative cervical lordosis of both groups had no statistical difference. Meanwhile, ADF group had a lower late neurological deterioration rate than LAMP group.

Dynamic factor can also affect clinical outcomes of LAMP. A larger postoperative C2–C7 range of motion was related to late neurological deterioration in patients with the segmental type of OPLL.22 The spontaneous fusion of vertebrae by the bridging of OPLL and the ossification of the anterior longitudinal ligament were considered as the most important factors protecting against deterioration in myelopathy.23 Goel et al24 found that treatment of OPLL with only fixation of the involved spinal segments without decompression had encouraging clinical outcomes. Fusion was advocated to conquer the postoperative kyphotic change and dynamic factor in the treatment of OPLL.5,24

To date, several researches that compared LAMP with laminectomy and fusion (LF) in the treatment of OPLL have been published. Chen et al7 reported that LF group had significantly higher recovery rate of JOA and postoperative JOA score than the LAMP group.7 Yuan et al25 reported that there was no statistically significant difference in the postoperative JOA scores between the two groups. Another paper did not report the outcomes of neurological function, but it did report that the LAMP group had a significantly higher incidence of progression of OPLL than the LF group.26 More studies are needed to compare LAMP and LF with regard to clinical outcomes in the treatment of OPLL and to further evaluate the impacts of fusion.

Postoperative progression of OPLL has been reported in both LAMP and ADF. Iwasaki et al27 reported that postoperative progression of OPLL after LAMP was observed in 70% of the patients over a 10-year follow-up. Another team, Matsuoka et al,28 reported that a marked postoperative progression of the OPLL after ADF was observed in 16.7% of the patients over a 10-year follow-up. Sakai et al8 reported that postoperative progression of the OPLL at 5-year follow-up period was observed in 5.0% of the ADF group and in 50.0% of the LAMP group. Progression of OPLL, loss of cervical lordosis, and dynamic factor may be the main causes of late neurological deterioration of LAMP in the treatment of OPLL. Insufficient decompression was more associated with late neurological deterioration in ADF group.23

In our meta-analysis, the mean incidence of late neurological deterioration in ADF and LAMP groups was 1.9% and 11.1%, respectively. It indicated that ADF could reduce the incidence of late neurological deterioration than LAMP in the treatment of OPLL.

Nowadays, many new techniques and methods are applied to remove the OPLL and evaluate the risks in anterior approach.29,30 No statistically significant difference was observed with regard to the complication rate of both groups in our meta-analysis. However, the reoperation rate was significantly higher in ADF group than in the LAMP group in our meta-analysis. Dural tear, graft extrusion, pseudarthrosis, and iatrogenic neurological deterioration were more often reported in ADF group, while C5 palsy and axial pain were also reported more frequently in the LAMP group. In four papers included in our analysis, the causes of reoperation were graft extrusion (6.8%), pseudarthrosis (4.1%), insufficient decompression (4.1%), late neurological deterioration (2.7%), C5 palsy (1.4%), and hematoma (1.4%) in ADF group (20.5%) and hematoma (1.7%), progression of OPLL (0.9%), and late neurological deterioration (0.9%) in the LAMP group (3.5%), respectively. The reoperation rate of ADF group was almost six times that of LAMP group. Graft extrusion and pseudarthrosis were the main causes of reoperation in ADF group. In these studies, the mean surgical level of ADF was from 3.0 to 3.3. Multilevel cervical corpectomy was associated with increased graft-related complications compared with single-level procedure.31 Biomechanical analysis suggested that stability of long-segment anterior plate fixation after a multilevel (≥2) corpectomy was insufficient.31,32 The cantilever force generated at the screw–bone interface increased as the length of the fused segment or plate length increased. As a result, anterior plate fixation tended to fail at the caudal end of the construct in multilevel corpectomy.32 Combined anterior–posterior fixation or posterior-only fixation was recommended in multilevel cervical corpectomy, which was significantly more rigid than the anterior-only fixation.33,34 However, in the studies discussed here, only anterior fixation was performed in ADF group, which might play an important role in graft extrusion and pseudarthrosis. Insufficient decompression was another main cause of reoperation in ADF group. Continuous-type OPLL, ossification of the dura, massive bleeding from the epidural space, and technical difficulties were the possible reasons for insufficient decompression in ADF group in the treatment of cervical OPLL.6,10,12

Meanwhile, there are some limitations to this meta-analysis. First, the included publications are in English; thus, a potential language bias may exist in this meta-analysis. Second, the sample size may not be large enough to find the possible existing evidence; therefore, larger-scale and higher-quality studies are needed to provide more reliable evidence for future evaluation. Third, there was a variable length of time in the follow-ups between some of the studies, and this complicated the evaluation and comparison of the surgical results. Fourth, clinical heterogeneity may be caused by the various indications for surgery and for the use of certain surgical technologies at the different treatment centers. Finally, the included patients with OPLL had a high mean occupation ratio (>50%). Some studies reported that the low occupation ratio group had significantly better clinical outcomes than high occupation ratio group (>50% or 60%) in the treatment of cervical OPLL with LAMP;8,10 therefore, our conclusion was based on a high mean occupation ratio. Further studies are needed regarding comparison of ADF versus LAMP in the treatment of cervical OPLL with a low mean occupation ratio.

Conclusion

In this meta-analysis, we systematically compared ADF with LAMP with regard to the clinical outcomes, complications, reoperation rates, and late neurological deterioration between ADF and LAMP in the treatment of multilevel cervical OPLL. ADF had better postoperative neurological function, neurological recovery rate, and less late neurological deterioration than LAMP in the treatment of cervical OPLL, with a high mean occupation ratio. LAMP had a decreased postoperative cervical lordosis that might be a cause of late neurological deterioration. The complication rates of both groups had no statistical difference. However, the reoperation rate was significantly higher in ADF group compared with LAMP group. The reoperation rate of ADF group was almost six times that of LAMP group. Benefits and risks should be balanced when ADF or LAMP is selected.