Risk factors and consequences of conversion in minimally invasive distal pancreatectomy.

Background: Although recent studies have reported potential benefits of laparoscopic approach in distal pancreatectomy, reports of conversion during minimally invasive distal pancreatectomy (MIDP) were limited.
Methods: This was a retrospective study using data from Sir Run Run Shaw Hospital around May 2013 to December 2018. Outcomes of patients who had conversions during MIDP were compared with patients with successful MIDP and with patients undergoing open distal pancreatectomy (ODP).
Results: Two-hundred and eighty-three cases were included in this study: 225 (79.5%) had MIDP, 30 (10.6%) had conversions and 28 (9.9%) had outpatient department. The risk factors for conversion included large lesion size (heart rates [HR]: 5.632, 95% confidencevinterval [CI]: 1.036-1.450, P = 0.018) and pancreatic cancer (HR: 6.957, 95% CI: 1.359-8.022, P = 0.009). Compared with MIDP, those who required conversion were associated with longer operations (P = 0.003), higher blood loss (P < 0.001) and more severe of the complications (P < 0.001). However, no statistically significant differences were found between the conversion group and ODP. Conclusions: Large lesion size and pancreatic cancer were reported to be independent risk factors for conversion during MIDP. As for post-operative outcomes, the outcomes of successfully MIDP were better than those for conversion. However, conversion did not lead to worsening outcomes when compared with ODP.

INTRODUCTION

Minimally invasive distal pancreatectomy (MIDP), which includes laparoscopic distal pancreatectomy (LDP) and robotic-assisted surgery, is a proven and safe method. In 1994, Cuschieri[1] were the first to demonstrate MIDP, and its usage rate has increased in recent years.[23] Some systematic reviews of cohort studies supported minimally invasive methods, reducing intraoperative blood loss, blood transfusion, rate of wound infection and complications, shortening hospital stays and increasing spleen preservation rates.[34] Parameters of oncology, such as range of resection and lymph node dissection, showed that MIDP was not worse than open distal pancreatectomy (ODP), especially in high-volume centres.[56]

The probability of MIDP conversion to ODP is 0%–30%.[7] There are many reasons for the transformation, including tumour invasion into the surrounding organs, bleeding, adhesions, inadequate visual field exposure, proximity to major blood vessels, assessment of the edges and preservation of the spleen. However, one study has shown that with the proficiency and standardisation of treatment methods, conversion rates in patients with MIDP can be reduced. At the same time, the probability of robotic collaboration is less than that of LDP.[8] Another survey reflected that many surgeons believed that MIDP was ultimately beneficial to applicable patients.[9] Controversially, switching to ODP during MIDP should not be regard as a complication, but little is known about the prognosis of patients who were converted.[10]

Hence, the purpose of this study is to compare three groups of patients (patients who had conversions during MIDP, patients with successfully completed MIDP and patients undergoing ODP) to analyse the risk factors and post-operative consequences of conversion in patients who have had MIDP.

METHODS

Patient selection

The clinical data of 225 cases of MIDP, 30 cases of conversions and 28 cases of ODP performed at Sir Run Run Shaw Hospital from May 2013 to December 2018 were used in this study. The study was approved by an institutional ethical board. The MIDP group included patients who underwent LDP and robotic distal pancreatectomy. Patients with pancreaticoenterostomy, simultaneous resection for abdominal organs (except spleen) and emergency surgery were excluded from the study. Figure 1a provides a flow chart for the study.

Figure 1

A flow chart for the study (a), the evolution in the number of patients who had conversion per year (b), the most frequent reason for conversion (c)

Study design

Risk factors for conversion were analysed by patient-specific and perioperative factors, including demographic data, comorbidities, surgical history, tumour characteristics and tumour marker levels. Perioperative factors that differed significantly on univariate analysis (P < 0.10) were included into a multivariable analysis to analyse risk factors. Table 1 shows n 6:1 propensity score matching in conversions versus MIDP and Table 2 shows a 1:1 propensity score matching in conversions versus ODP based on the patients’ age, gender, body mass index, America Society of Anaesthesiologists fitness grade, a current habit of smoking and drinking, comorbidities (hypertension, diabetes mellitus, heart disease and lung disease), surgical history, tumour characteristics and tumour marker levels. Patients who had conversion were compared with successfully MIDPs (laparoscopic or robotic) and with patients who had ODP. The outcomes collected included operative time, intraoperative blood loss, post-operative complications, pancreatic fistula, post-operative hospital stay, ICU occupancy, ICU check in time, 30- and 90-day mortality, 30-day readmission and hospitalisation costs.

Table 1

Comparison of the conversions group and the open distal pancreatectomy group: Demographic outcomes

Characteristics	Before propensity score matching			After propensity score matching
	Conversions (n=30)	ODP (n=28)	P	Conversions (n=28)	ODP (n=28)	P
Age (years)*	56.13±12.21	59.82±14.72	0.302	55.64±12.48	59.82±14.72	0.257
Gender (male/female)	11/19	13/15	0.626	10/18	13/15	0.587
BMI*	23±3.27	23.60±3.48	0.505	23.19±3.29	23.60±3.48	0.652
ASA fitness grade (%)
1	5 (16.7)	2 (7.1)	0.441	5 (17.9)	2 (7.1)	0.374
2	23 (76.7)	25 (89.3)	21 (75.0)	25 (89.3)
3	2 (6.7)	1 (3.6)	2 (7.1)	1 (3.6)
Hypertension (%)	8 (26.7)	11 (39.3)	0.457	8 (28.6)	11 (39.3)	0.572
Diabetes (%)	6 (20.0)	3 (10.7)	0.540	6 (21.4)	3 (10.7)	0.467
Tobacco use (%)	2 (6.7)	2 (7.1)	1.000	2 (7.1)	2 (7.1)	1.000
Alcohol (%)	0 (0.0)	2 (7.1)	0.441	0 (0.0)	2 (7.1)	0.471
Cardiovascular diseases (%)	0 (0.0)	0 (0.0)	NA	0 (0.0)	0 (0.0)	NA
Respiratory diseases (%)	0 (0.0)	0 (0.0)	NA	0 (0.0)	0 (0.0)	NA
Preoperative chemotherapy (%)	2 (6.7)	0 (0.0)	0.503	2 (7.1)	0 (0.0)	0.471
Previous abdominal surgery (%)	10 (33.3)	8 (28.6)	0.914	9 (32.1)	8 (28.6)	1.000
Lesion size*	5.05±2.43	4.51±2.90	0.450	5.14±2.47	4.51±2.90	0.389
Histology (%)
Benign tumour	5 (16.7)	4 (14.3)	0.791	5 (17.9)	4 (14.3)	0.744
Precancerous lesion	7 (23.3)	7 (25.0)		7 (25.0)	7 (25.0)
Malignant tumour	17 (56.7)	17 (60.7)		15 (53.6)	17 (60.7)
Pseudocyst	1 (3.3)	0 (0.0)		1 (3.6)	0 (0.0)
CA125*	31.94±48.44	43.44±141.38	0.687	29.00±46.03	43.44±141.38	0.623
CA199*	381.61±1069.34	222.12±432.90	0.482	199.10±389.44	222.12±432.90	0.841
CEA*	4.54±10.32	4.45±4.79	0.970	2.60±1.63	4.45±4.69	0.068

*Mean±SD. BMI: Body mass index, ASA: America Society of Anaesthesiologists, SD: Standard deviation, ODP: Open distal pancreatectomy

Table 2

Comparison of the conversions group and the minimally invasive distal pancreatectomy group: Demographic outcomes

Characteristics	Before propensity score matching			After propensity score matching
	Conversions (n=30)	MIDP (n=225)	P	Conversions (n=30)	MIDP (n=180)	P
Age*	56.13±12.21	51.87±15.09	0.139	56.13±12.21	52.05±15.03	0.159
Gender (male/female)	11/19	81/144	1.000	11/19	58/122	0.787
BMI*	23.00±3.27	22.90±3.11	0.863	23.00±3.27	22.87±3.12	0.833
ASA (%)
1	5 (16.7)	41 (18.2)	0.251	5 (16.7)	30 (16.7)	0.249
2	23 (76.7)	180 (80.0)		23 (76.7)	147 (81.7)
3	2 (6.7)	4 (1.8)		2 (6.7)	3 (1.7)
Hypertension (%)	8 (26.7)	64 (28.4)	1.000	8 (26.7)	46 (25.6)	1.000
Diabetes (%)	6 (20.0)	26 (11.6)	0.309	6 (20.0)	25 (13.9)	0.551
Tobacco use (%)	2 (6.7)	25 (11.1)	0.669	2 (6.7)	10 (5.6)	1.000
Alcohol (%)	0 (0.0)	21 (9.3)	0.164	30 (100.0)	180 (100.0)	NA
Cardiovascular diseases (%)	30 (100.0)	219 (97.3)	0.792	30 (100.0)	175 (97.2)	0.782
Respiratory diseases (%)	30 (100.0)	217 (96.4)	0.623	30 (100.0)	172 (95.6)	0.508
Preoperative chemotherapy (%)	0 (0.0)	3 (1.3)	1.000	0 (0.0)	3 (1.7)	1.000
Previous abdominal surgery (%)	10 (33.3)	57 (25.3)	0.475	10 (33.3)	51 (28.3)	0.733
Lesion size*	5.05±2.43	4.05±2.25	0.025	5.05±2.43	4.30±2.29	0.101
Histology (%)
Benign tumour	5 (16.7)	59 (26.2)	0.242	5 (16.7)	46 (25.6)	0.452
Precancerous lesion	7 (23.3)	73 (32.4)		7 (23.3)	52 (28.9)
Malignant tumour	17 (56.7)	84 (37.3)		17 (56.7)	74 (41.1)
Pseudocyst	1 (3.3)	9 (4.0)		1 (3.3)	8 (4.4)
CA125*	31.94±48.44	31.20±186.41	0.983	31.94±48.44	35.14±207.60	0.936
CA199*	381.61±1069.34	91.28±305.23	0.002	381.61±1069.34	93.18±315.47	0.004
CEA*	4.54±10.32	62.82±858.01	0.720	4.54±10.32	5.04±23.68

*Mean±SD. BMI: Body mass index, ASA: America Society of Anesthesiologists, SD: Standard deviation, MIDP: Minimally invasive distal pancreatectomy

Post-operative data

The post-operative complications are according to the Clavien–Dindo Classification grade standard. Grades 1–3a represent mild complications requiring drug therapy and interventional therapy, while Grades 3b–5 represent serious complications such as organ failure and even death, which requires a surgical approach.

A pancreatic fistula is when fluid amylase concentration is >3 times than that of serum amylase. On the other hand, pancreatic fistula B is when peripancreatic drainage is needed for more than 3 weeks and requires surgical interventions.

Statistical analysis

Quantitative variables were appropriately expressed as mean or median (range). Qualitative variables were presented as frequencies. The χ2 test or Fisher's exact test was used to compare categorical data, whereas the Mann–Whitney U-test was used for intergroup comparisons of quantitative variables. Multivariable logistic regression model was used to estimate the probabilities of experiencing post-operative complications and undergoing conversion. All variables that varied significantly in the univariable analysis (P < 0.100) between the two groups were included in the logistic model.

RESULTS

In this study, 283 cases were analysed, including 225 (79.5%) with MIDP, 30 (10.6%) with conversions and 28 (9.9%) with outpatient department (OPD). For the MIDP group, 210 (93.3%) were performed laparoscopically, while 15 (6.7%) were performed robotically. The overall conversion rate in this study was 15.3%. Figure 1b demonstrates the evolution in the number of patients who had conversion per year. The conversion rate dropped from 21% in 2013 to 10% in 2014, then kept stable from 2014 to 2018. Figure 1c summarises the reasons for conversion. Bleeding (n = 4, 13.3%) is difficult to control laparoscopically, but no massive acute bleeding from a major vascular structure occurred. The most frequent reason for conversion was failure of progress, with adhesion (n = 10, 33.3%) and anatomy of vessels (i.e., the vicinity of the pancreas including the celiac artery, portal vein and splenic artery and vein; n = 6, 20.0%), extent of tumour invasion (i.e., invasion to the surrounding visceral organs such as the duodenum, stomach and colon; n = 5, 16.7%) and venous invasion (n = 5, 16.7%).

Conversion versus minimally invasive distal pancreatectomy

In the analysis of outcome [Table 3], patients who required conversion were associated with longer operations (244.00 ± 76.32 vs. 194.15 ± 77.24 min; P = 0.003), higher intraoperative blood loss (591.54 ± 667.55 vs. 162.23 ± 193.56 ml; P < 0.001), a higher rate of transfusion (50% vs. 7.2%; P < 0.001) and a lower rate of spleen preserving (10.0% vs. 38.9%; P < 0.001) in the operation than those who had MIDP. After operation, the conversion group had a worse outcomes than the MIDP group, with a longer mean post-operative hospital stay (17.40 ± 10.03 vs. 11.15 ± 7.61 days; P < 0.001) and the severity of the complications (based on Clavien–Dindo classification) in conversion was significantly higher in the MIDP group (P < 0.001). In the two groups, the proportion of patients with post-operative pancreatic fistula was similar. In addition, conversion cases paid higher costs (82,990.17 ± 43,101.39 vs. 65,419.43 ± 34,094.03; P = 0.013), whereas ICU occupancy rate, ICU check in time, 30-day morbidity rate, 90-day mortality rates and 30-day readmission rate were comparable (P = 0.162, P = 0.156, P = 0.306, P = 1.000 and P = 0.892, respectively).

Table 3

Comparison of the conversions group, open distal pancreatectomy group and the minimally invasive distal pancreatectomy group: Post-operative outcomes and prognosis

	MIDP (n=180)	Conversions (n=30)	P	ODP (n=28)	Conversions (n=28)	P
Spleen preserving (%)	70 (38.9)	3 (10.0)	<0.001	3 (10.7)	3 (10.7)	1.000
Blood loss*	162.23±193.56	591.54±667.55	<0.001	379.63±280.88	578.33±690.34	0.176
Transfusion (%)	13 (7.2)	15 (50.0)	<0.001	11 (39.3)	14 (50)	0.591
Operative time*	194.15±77.24	244.00±76.32	0.003	226.27±85.80	242.39±79.41	0.500
Complications (%)
0	66 (36.7)	3 (10.0)	<0.001	2 (7.1)	3 (10.7)	0.332
1	53 (29.4)	4 (13.3)		10 (35.7)	3 (10.7)
2	39 (21.7)	17 (56.7)		12 (42.9)	16 (57.1)
3a	14 (7.8)	3 (10.0)		2 (7.1)	3 (10.7)
3b	2 (1.1)	0 (0.0)		0 (0.0)	0 (0.0)
4a	5 (2.8)	2 (6.7)		2 (7.1)	2 (7.1)
4b	1 (0.6)	0 (0.0)		0 (0.0)	0 (0.0)
5	0 (0.0)	1 (3.3)		0 (0.0)	1 (3.6)
Pancreatic fistula (%)
B	44 (24.4)	12 (40.0)	0.191	7 (25.0)	11 (39.3)	0.277
C	5 (2.8)	1 (3.3)		0 (0.0)	1 (3.6)
N	131 (72.8)	17 (56.7)		21 (75.0)	16 (57.1)
ICU (%)	5 (2.8)	3 (10.0)	0.162	2 (7.1)	3 (10.7)	1.000
ICU stay*	0.17±1.40	0.63±2.77	0.156	0.14±0.52	0.68±2.87	0.335
Hospital stay*	11.15±7.61	17.40±10.03	<0.001	14.11±7.50	17.54±10.38	0.162
X30-day mortality (%)	0 (0.0)	1 (3.3)	0.306	0 (0.0)	1 (3.6)	1.000
X90-day mortality (%)	1 (0.6)	0 (0.0)	1.000	0 (0.0)	0 (0.0)	1.000
X30-day readmission (%)	19 (10.6)	4 (13.3)	0.892	2 (7.1)	4 (14.3)	0.666
Costs*	65,419.43±34,094.03	82,990.17±43,101.39	0.013	63,517.54±30,065.88	83,288.76±44,612.93	0.057

*Mean±SD. SD: Standard deviation, ODP: Open distal pancreatectomy, MIDP: Minimally invasive distal pancreatectomy

Conversion versus open

Operations time, intraoperative blood loss, rate of transfusion and the rate of spleen preserving were observed at the surgery of 56 patients [Table 3]. However, no statistically significant differences were found between the conversion group and open group. At the same time, the mean post-operative hospital stay (17.54 ± 10.38 vs. 14.11 ± 7.50 days; P = 0.162) was comparable between the two groups. There was no statistical differences in short-term post-operative outcome, with equal rates and severity of the complications (P = 0.332) and equal proportion of patients with post-operative pancreatic fistula (P = 0.227). ICU occupancy rate, ICU check in time, 30-day morbidity rate, 90-day mortality rate and 30-day readmission rate were also similar (P = 1.000, P = 0.335, P = 1.000, P = 1.000 and P = 0.666, respectively). The mean medical cost of the conversion group was higher than the MIDP but not significantly greater. The risk factors for conversion included large lesion size (heart rates [HR]: 5.632, 95% CI: 1.036–1.450, P = 0.018) and pancreatic cancer (HR: 6.957, 95% CI: 1.359–8.022, P = 0.009) [Table 4].

Table 4

Factors associated with conversion

	OR (95% CI)	P
Pancreatic cancer	6.957 (1.359-8.022)	0.008
Lesion size	5.632 (1.036-1.450)	0.018
CA199	1.709 (1.000-1.001)	0.191

OR: Odds ratio, CI: Confidence interval

DISCUSSION

In this study, we included 283 patients, including MIDP (225), conversion (30) and OPD (28), to analyse for risk factors of conversion and the influence of conversion on the post-operative outcomes. Conversion during MIDP occurred in 11.7% patients. Large lesion size and pancreatic cancer were determined to be the high-risk factors of conversion. We identified that conversion was worse than successfully MIDP in regard to post-operative outcomes. However, the outcomes of conversion and ODP were quite similar.

In the present study, the conversion rate dropped from 21% in 2013 to 10% in 2014 and remained stable from 2014 to 2018. This proves the frequent occurrence of conversion during the early experience of MIDP. MIDP is difficult in the early learning curve, even for surgeons who are familiar with laparoscopic surgery or in some high-volume centres. De Rooij et al.[11] reported that the conversion rate after distal pancreatectomy was 38% when done by an inexperienced surgeon, and the conversion rate dropped to 8% after professional training. Systematic training for MIDP is essential for reducing the conversion rate in the early learning curve. The explanation for the stabilisation from 2014 to 2018 may be due to two reasons. First, with a rapid development of laparoscopic technology and dramatically extended indications for laparoscopy over the past 10 years, surgeons in our centre have dared to take on more challenging cases and more patients with large size lesions, allowing patients with pancreatic cancer a chance at laparoscopic surgery. In these patients, the tumour may sometimes have invaded surrounding tissues, and the exposure and anatomy of vessels can be difficult, which increases the risk of conversion. Second, with the limitation of laparoscopic technology, the rate of conversion may reach its lowest rate. Comparing 391 LDP and 246 robotic cases, a recent meta-analysis showed that the rate of conversion was higher in the LDP group (20%) than in the robotic group (9%) (odds ratio [OR]: 0.61; 95% confidence interval [CI]: 0.36–1.05; P = 0.08).[12] We believe that the increasing usage of robotic surgery will continue to reduce the rate of conversion even further.

The majority of the studies about MIDP largely focus on the short-term and oncologic outcomes after surgery of pancreatic cancer, especially when compared with ODP. However, most of these studies fail to discuss the impact of conversion. In regard to the impact of conversion in MIDP, the question of whether conversion leads to worse post-operative outcomes is seldom reported. It has been reported in only two studies that the patients who underwent conversion after MIDP had a worse post-operative outcomes than complete MIDP.[213] In this study, the conversion group was associated with longer operations time, higher intraoperative blood loss, a higher rate of transfusion, a lower rate of spleen preserving and worse post-operative outcome than those who had successfully MIDP. Thus, reducing the rate of conversions is quite important. Conversion rate can decrease if MIDP indications are controlled strictly, operative skills are improved and the robotic technology can be developed. In this study, while the conversion group was associated with worse post-operative outcomes than successfully MIDP but equal post-operative outcomes than ODP. It demonstrates that though conversion, the benefits of MIDP are offset and do not bring worse outcomes than ODP. This study raises the question of “why not start with MIDP for all patients?” Even if the procedure of MIDP fails, there are no worse post-operative outcomes for conversion to ODP. If the procedure of MIDP is a success, the patients can receive the benefit of laparoscopic techniques. This leads to a strong argument against patients who have the high risk of conversion that should require ODP. However, the two studies[213] mentioned above showed that the outcomes for patients experiencing conversion were worse or no better than planned open resection. The explanation for this difference may be due to two reasons. First, the patients’ data in these two studies were collected from 120 hospitals. The inclusion criteria of patients and medical level of hospitals varied and most of the hospitals may have been from low-volume centres. However, patients with pancreaticoenterostomy, simultaneous resection for abdominal organs (except spleen) and emergency surgery were excluded in our centre, which removed interference from various inclusion criteria of patients. The laparoscopic techniques in our centre, as a high-volume centre, are considered top in the country, which makes the conversion procedure relatively safer. Second, the determination of adequate conversion time was helpful to the outcomes after the conversions. In this study, the main reason of conversion was a failure of the procedure (86.7%), including adhesions, anatomy of vessels and extent of tumour invasion. However, in the two studies, bleeding is the most common causes. Complications during surgery such as bleeding can lead to post-operative adverse consequences. To some extent, active conversion was equal to the sum of laparoscopic exploration and laparotomy. Thus, we should not regard conversion as an intraoperative complication but as an indicator of the surgeons’ ability to determine adequate conversion time to avoid adverse outcomes after conversion. There should be a process of actively seeking the best surgical plan for the patients’ condition, rather than passive conversion due to major bleeding and other reasons.

A study reported by Hua et al.[14] identified malignant disease as a risk factor for conversion of MIDP. In this study, however, pancreatic cancer, rather than the malignant tumour, was considered to be an independent risk factor of conversion. Large lesion size represented tumour progression to some extent. We believe that the invasion of tumour to the surrounding tissue and vessels makes dissection difficult, leading to the failure of laparoscope procedure and an increased risk of injury to vessels.

Pre-operative testing revealed CA199 levels to be significantly higher in the conversion group, which may be related to pancreatic cancer and lead to tumour invasion, pancreatitis and severe adhesions. The use of computed tomography for the evaluation of the tumour stage and pathologic type could help surgeons to predict the difficulty of dissection in the surgery and the risk of conversion. These evaluations would help to determine the choice of operative methods in low-volume centres and adequate conversion time in high-volume centres. The use of CA199 to predict the risk of conversion requires more studies for verification.

This study has several limitations. First, the study's retrospective nature and relatively small number of cases might lead to insufficient statistical power in the analysis of the outcomes of conversion and a possibility of a type II error. Second, as a single-centre study, especially in a high-volume centre, our study does not represent the average medical level.

Concluson

Large lesion size and pancreatic cancer were reported to be independent risk factors for conversion during MIDP. As for post-operative outcomes, the outcomes of successfully MIDP were better than those for conversion. However, conversion did not lead to worsening outcomes when compared with ODP. Our data suggest that MIDP should be encouraged in developed minimally invasive surgery center without worry about the risks of conversion.

Author contribution

L-P, B-Z, S-J and S-F substantial contributed to conception and design, acquisition of data or analysis and interpretation of data and Z-YJ, M-YC and X-JC drafted the article or revise it critically for important intellectual content, all the authors finally approved the version to be published.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.