Hidden Blood Loss in Percutaneous Endoscopic Lumbar Discectomy (PELD): a Prospective Study

Background: Percutaneous endoscopic lumbar discectomy (PELD) is a minimally invasive technique for treatment of lumbar disc herniation (LDH) that requires only an eight-mm skin incision and promotes faster recovery. It is widely believed that the procedure is associated with minimal blood loss. However, signicant perioperative hidden blood loss (HBL) is frequently unaccounted for. This study aimed to investigate HBL and peri-operative factors contributing to HBL in a series of individuals undergoing PELD. Methods: As a prospective observational study, 156 patients, mean age 41.6 years (ranged from 17 to 71), undergoing PELD were nally enrolled between May 2019 and November 2020. The analyzed factors included gender, age, body mass index (BMI), symptom duration, operation approach/technique, operation duration, the presence of associated chronic diseases (e.g., hypertension, rheumatoid arthritis, and diabetes mellitus), and improvements in Visual Analog Scale (VAS), Japanese Orthopedic Association (JOA), and Oswestry Disability Index (ODI) scores. Gross’s formula was applied to calculate blood loss from recorded values for patient height, weight, and hematocrit levels before and after surgery. Multivariate linear regression analysis and t test were performed to identify factors that contributed signicantly to HBL. Results: The mean HBL during PELD was 201 ± 126 mL, and the Hb reduction observed post-operatively was 7.4 ± 4.5 g/L. A lateral surgical approach was associated with greater HBL compared with an interlaminar approach. However, no signicant differences in VAS, JOA, and ODI score improvements were noted between the two surgical approaches. Increased surgical times and foraminal decompression were identied by multivariate linear regression analysis as linked to increased HBL. As expected, the occurrence of post-operative anemia was signicantly associated with HBL. Age, gender, BMI, symptom duration, and associated chronic diseases showed no association with increased HBL during PELD. Conclusions: HBL is increased in PELD patients with long surgical times and lumbar foraminal decompression.


Introduction
As a minimally invasive surgical procedure, percutaneous endoscopic lumbar discectomy (PELD) offers a potentially improved treatment approach for lumber disc herniation (LDH) and some types of lumbar spinal stenosis (LSS) 1 . With several studies reporting encouraging results, the procedure has several advantages compared to other surgical methods, including a reduced risk of damage to the posterior and paraspinal structures, which further reduces the induced trauma compared with traditional open surgery, thereby also reducing intraoperative blood and promoting more rapid postoperative recovery [2][3][4][5] . When performed by an experienced surgeon in clinical practice, signi cant intraoperative bleeding during PELD is rarely observed, with the estimated blood loss being less than 50 ml 6 .
Hidden blood loss (HBL) is an important factor in surgical treatments. Brecher et al 7 devised a mathematical model for calculating intraoperative blood loss that takes into account the patient's blood volume (PBV), both the amount and type of red blood cells transfused, the patient's hematocrit (Hct) levels preoperatively and at discharge, the volume of salvaged blood transfused during the procedure, and the amount of hemodilution carried out during surgery. The amount of blood loss calculated by this model is on average 2-fold greater than that estimated by the traditional intraoperative approach for all surgeries. To date, the amount of HBL in PELD and the underlying causes have not been reported. A clear understanding of HBL during PELD and its causes would be valuable for improving this technique and its clinical outcomes.
Therefore, we hypothesized that there may be unappreciated HBL during spinal surgery and performed the present prospective observational study among patients who underwent PELD at our institution and to investigate this possibility and the likely causes. From the patients' clinical data, we calculated the HBL and analyzed several possible risk factors associated with increased bleeding in PELD.

Patients
The protocol for this study was approved by the Ethics Committee of Guangdong Provincial People's Hospital. We retrieved the electronic medical records of 323 patients underwent surgical treatment by PELD between May 2019 and November 2020. Patients were excluded according to the following criteria: 1) hematological diseases; 2) previous lumber surgery or trauma and transfusion history including blood products and crystalloids; 3) abnormal blood coagulation indexes or use of antiplatelet and anticoagulant drugs or herbal medications within 1 week of surgery; 4) abuse of tobacco (≥ 20 cigarettes/day), alcohol (liquor ≥ 90 ml/day or 150 ml/week), or recreational drugs; and 5) current menstruation. The study followed a sample-of-convenience design that included all identi ed patients who met the inclusion/exclusion criteria.
For each patient, we recorded gender, age, weight, height, body mass index (BMI), hematocrit (Hct) levels before and after surgery, hemoglobin (Hb) levels before and after surgery, coagulation function indexes, the levels at which LHD and LSS were applied, the surgical approach, and comorbidities (e.g., hypertension, rheumatoid arthritis, and diabetes mellitus). The pre-operative durations of symptoms and characteristics were also recorded. All patients received treatment with regular non-surgical techniques or medications (e.g., functional exercise, non-steroidal anti-in ammatory drugs [NSAIDs], and vitamins B 1 and B 12 ) in local or community hospitals prior to surgical therapy, with symptoms continuing to affect their daily life. Two rheumatoid arthritis patients in our cohort accepted PELD and stopped steroid use before surgery. Back and leg pain as well as improvements in daily function were assessed by comparing visual analog scale (VAS), Japanese Orthopedic Association (JOA), and Oswestry Disability Index (ODI) scores on postoperative day 1 versus to preoperative values. No antibiotics were administered during the peri-operative period.

Surgical procedures
All PELD surgeries were performed by senior surgeons. The method used for anesthesia was decided with input from the patients and/or their relatives. The surgery was performed with the patient lying in prone position on a carbon-ber operating bed, so that X-ray images could be taking during the surgery. A transforaminal approach was used for cases of L4/5 disease and an interlaminar approach was applied for treatment of L5/S1 lesions. An incision approximately 8 mm long was made in the skin. Then through the guide needle, a working channel was inserted into the posterior wall of the spinal canal along with an expansion cannula, through which the operating system was deployed. The epidural fat, nerve root, disc avum ligament space, herniated disc tissue, and other spinal canal structures were viewed by endoscopy. One typical case has been illustrated in Fig. 1 for simply discectomy and if necessary, percutaneous endoscopic lumbar foraminoplasty (PELF) or percutaneous endoscopic ventral facetectomy (PEVF) was performed with a high-speed drill (Fig. 2) and/or trephine reamer (Fig. 3) for patients with foraminal or lateral recess stenosis undergoing LSS 8,9 . After complete hemostasis was achieved by the bipolar technique, the channel and light source were slowly removed. Finally, the skin incision was sutured with one stitch. During the surgery, the core body temperature was monitored and maintained within the normal range.

Post-operative care
For all patients, complete blood count analysis, including Hct, was performed before the operation and on post-operative day 1. Post-operative day 1 was chosen because by this day, the patients were hemodynamically stable and no further uid shifts were likely. Anemia was de ned based on the measured hemoglobin (Hb) levels according to the World Health Organization thresholds for women (< 120 g/L) and men (< 130 g/L) 10 . Oral NSAIDs and vitamin B 1 /B 12 were given post-operatively. Patients were discharged 1 or 2 days after surgery.

HBL calculation
The total patient's blood volume (PBV) was calculated as described by Nadler et al. 11 using the following formula: PBV = k1 × height (m) 3 + k2×weight (kg) + k3. For male patients, k1 = 0.3669, k2 = 0.03219, and k3 = 0.6041, and for female patients, k1 = 0.3561, k2 = 0.03308, and k3 = 0.1833. The Gross formula was applied for calculation of the total blood loss (TBL) which equal to HBL in the perioperative period 12 : TBL = HBL = PBV × (Hct pre −Hct post )/Hct ave , where Hct pre is the preoperative Hct level, Hct post is the Hct level on post-operative day 1, and Hct ave is the average of the Hct pre and Hct post levels. In addition, the Hct method has been identi ed as a reliable way to estimate HBL 13 .
Statistical analysis SPSS version 24.0 (SPSS, Inc., Chicago, IL, USA) was used for all statistical analyses. Differences according to sex or surgical approach were evaluated using independent-samples Student's t tests Differences in anemic status from before to after surgery were identi ed using the Chi-square test. The Pearson or Spearman method was used to text for correlation with HBL. Multivariable linear regression analysis was conducted to determine which peri-operative factors were independently associated with HBL from among four quantitative variables (i.e., age, BMI, surgical time, duration of symptoms) and six qualitative variables (i.e., surgical approach, symptomatic side, decompression procedures, hypertension, rheumatoid arthritis, and diabetes mellitus). For qualitative variables, the transforaminal approach, leftside symptoms, decompression, hypertension, rheumatoid arthritis, and diabetes mellitus were designated as ''1''. For the interlaminar approach, right-side symptoms, non-decompression, nonhypertension, non-rheumatoid arthritis, or non-diabetes mellitus were designated as ''0''. A positive value for the coe cient indicated that the variable positively in uenced HBL (the dependent variable), whereas a negative value for the coe cient indicated that the variable negatively in uenced HBL. The ''Enter'' method was applied to incorporate all variables in the model. P values < 0.05 indicated statistical signi cance.

Demographic Data
From May 2019 to November 2020, 323 patients underwent PELD, and 156 of these patients met the inclusion/exclusion criteria and were included in the present study. The demographic and clinical characteristics of the patients are presented in Table 1. The patients ranged in age from 17 to 71 years, and the mean age for all patients in the study was 41.6 years. The pre-operative duration of symptoms in these patients ranged from 6 to 55 months (22.9 ± 12.4 months). Clinical Outcomes No blood products were given to any patient during the study, and no patients required wound drainage. Because essentially no blood loss was noted during the PELD procedure for any patient, visible blood loss was disregarded and HBL was considered a good approximation of total blood loss. The calculated values for percent reduction in Hct level, percent reduction in Hb level, level of Hb loss, duration of symptoms, and HBL are presented in Table 1. For all patients, the mean Hb loss was 7.4 ± 4.5 g/L, and the mean HBL was 201 ± 126 mL. The mean HBL did not differ signi cantly between male and female patients (206 ± 127 mL vs. 216 ± 124 mL, P = 0.641). However, the mean HBL for PELD via the lateral approach (L4/5) was signi cantly greater than that for PELD via the interlaminar approach (L5/S1) (228 ± 126 mL vs. 179 ± 121 mL, P = 0.02). No signi cant differences (all P > 0.05) were observed between the two surgical types in terms of the improvements in VAS (5.34 vs 4.68), JOA (77.2% vs 76.2%), or ODI (16.9% vs 18.3%). The 25th -percentile and 75th -percentile cases for HBL were compared, and the details are presented in Table 2. No complication has been observed in this cohort. Contributing factors to HBL HBL increased relatively proportionately (R 2 = 0.6545) with increasing surgical time (Fig. 4), and the comparison of pre-operative to post-operative anemic status showed that HBL was correlated with an increased incidence of post-operative anemia (P = 0.000, Chi-square test; Table 3). Correlation analysis revealed signi cant correlations of increased HBL with surgical time (P = 0.001), decompression (P = 0.001), and surgical approach (P = 0.015) ( Table 4). Next, multivariable linear regression analysis identi ed surgical time (P = 0.000) and decompression (P = 0.001) as positively associated with HBL (Table 5).

Discussion
As a minimally invasive technique for treatment of LDH that PELD only requires an eight-mm skin incision and promotes faster recovery for patients. Previously, it has been widely believed that PELD associated with minimal blood loss based on the intra-operative observation. However, signi cant perioperative HBL is frequently unaccounted for.
In 2000, Sehat et al 14 gave further support to the concept of HBL when they reported that HBL during total hip replacement represented 49% of the total blood loss. Recently, HBL has received increased attention among spine surgeons. HBL was shown to constitute approximately 40% of total blood loss during primary and revision posterior spinal fusion surgeries 15 . Wen et al 16 reported that signi cant HBL occurred in patients who underwent posterior lumber fusion surgery for degeneration, especially in cases of multi-level fusion. An additional study concluded that HBL was underestimated and accounted for a large percentage of total blood loss in minimally invasive transforaminal lumbar interbody fusion 17 . HBL was calculated to range from 678 to 1,267 mL in two-or three-level posterior lumbar decompression and fusion 18 , suggesting the need to consider HBL in patient management. Furthermore, for patients with rheumatoid arthritis undergoing posterior lumbar interbody fusion, while total blood loss, intraoperative bleeding, and operation time showed no variation, HBL was greater than that for patients without rheumatoid arthritis, particularly for long-segmental surgery 19 . Despite these ndings, HBL during PELD is not widely recognized and has not been characterized in the literature. To the best of our knowledge, it is the rst prospective observational study of HBL in patients undergoing PELD.
Percutaneous kyphoplasty (PKP) is also a minimally invasive procedure used to treat osteoporotic vertebral compression fractures without drainage, and HBL during PKP is di cult to estimate. Wu et al 20 concluded that HBL should not be ignored during the peri-operative period for vertebroplasty, especially in patients who are frail and/or have multiple fractures and have PKP. However, the causes of PELDassociated HBL are still unknown. Several possible sources of HBL in PELD include hemolysis, extravasation of blood into surrounding tissues, residual blood in a dead space, or simple underestimation of blood loss.
The concept of percutaneous posterolateral nucleotomy was introduced in 1973, and PELD was subsequently applied for the treatment of lumbar disc herniation 21 . PELD refers to both percutaneous endoscopic transforaminal discectomy (PETD) and percutaneous endoscopic interlaminar discectomy (PEID). As a minimally invasive spinal procedure, PELD has gained popularity due to its small incision, quick recovery, short hospitalization, and equivalent clinical outcomes compared to open surgery 22 . In the present study, all patients were discharged within 1-2 days post-operation.
The transforaminal approach is the procedure used most commonly to treat most types of herniated nucleus pulposus, except for the intra-canalicular type at L5/S1 with a high iliac crest 9 . Herein, we found that the lateral approach was associated with approximately 50 mL more HBL than the interlaminar approach. The improvements in VAS, JOA, and ODI scores on post-operative day 1 were similar with both surgical approaches. While one study reported no signi cant differences in estimated blood loss, post-operative periods in bed or hospitalization time between PETD and PEID 23 , there remains a lack of su cient evidence to compare these two approaches. In the present study, a transforaminal approach was elected for treating L4/L5, and an interlaminar approach for treating L5/S1, based on the hypothesis that a transforaminal approach would require cutting through more muscle and induce more trauma. However, our multivariable linear regression analysis showed that the surgical approach in PELD was not associated with the amount of HBL.
Our analysis did reveal a positive correlation between increased HBL and increased surgical time. Greater soft tissue damage may have partially accounted for this correlation. Generally, the greatest sources of HBL during surgery have been considered to extravasation of blood into surrounding tissues and hemolysis 24,25 . In patients who receive treatment for LSS and also undergo foraminoplasty or ventral facetectomy, high-speed drilling and/or trephine reamers are applied, which increases bone-derived bleeding. In the present study, we found that application of decompression procedures was positively correlated with HBL.
The present study has several limitations. The cohort size was small, and control groups were not included. Additionally, the data analysis was not done in a blinded manner. Moreover, post-operative Hct and Hb levels were measured only on post-operative day 1. In future studies, these levels will need to be measured at later time points in order to analyze the return to normal levels. Although core body temperature was monitored and maintained during surgery, a large amount of uid ushing and other hypothermia factors were di cult to avoid. Fluid shifts had likely already occurred in these patients, which may have confounded Hct estimation.

Conclusions
The results of the current study indicate that HBL is much larger in PELD than previous expectations, which based on the intra-operative observation. The performance of additional decompression procedures and increased surgical times appear to increase PELD-related HBL. Other factors, such as age and BMI, did not in uence HBL in this study. Overall, our ndings emphasize the need to consider HBL in PELD, especially when planning surgical treatment for elderly frail patients.

Consent for publication
Patients signed informed consent regarding publishing their data and photographs.

Availability of data and materials
The datasets used and analysed during the current study are available from the corresponding author on reasonable request.
CC, YBC, and HLG contributed to the conception of the study. CC and DX wrote the manuscript. GYL, RYZ, XQZ, and CXL reviewed the manuscript. All authors read and approved the nal manuscript.   showing L5/S1 disc herniation. (C) Partial ventral facetectomy was performed using a trephine reamer.