Safety of Intraoperative Cell Salvage in Cancer Surgery: An Updated Meta-Analysis of the Current Literature

Perioperative allogeneic red blood cell (RBC) transfusion is associated with increased mortality and cancer recurrence in multiple retrospective studies [1-6]. The underlying trials adjusted their empirical observations for confounding factors such as preoperative anemia, severity of illness, perioperative blood loss, variations in hemotherapy algorithms, and extent of surgical trauma [7]. Some results suggest that transfusion-related immunomodulation may affect transfusion-associated outcome results [8]. Although the controversy continues whether the observed outcome after allogeneic RBC transfusion is due to correlation or causation, blood-sparing techniques are important to reduce transfusion-associated adverse events [9, 10].

Blood-sparing techniques, especially in cancer surgery, should be investigated with regards to their intrinsic safety profile [10]. Intraoperative autologous blood cell salvage (IOCS) is one method to reduce the rate and amount of allogeneic RBC transfusion [11]. The principle of IOCS is to collect and process blood from the surgical site and reinfuse the autologous blood. IOCS has been shown to significantly reduce allogeneic transfusion in metastatic spine surgery as well as adult cardiac and orthopedic surgery [11, 12]. The modern IOCS nowadays uses autologous blood return following filtration with a leukocyte reduction filter (“LRF” or leukocyte depletion “LDF” [“depletion” is defined as a more stringent reduction of leucocytes to less than 10⁶ per unit]) in order to reduce contamination from tumor cells. Despite this additional safety step, the safety of IOCS in cancer surgery is controversial, and we had yet to see prospective randomized controlled trials on this topic.

Two questions need to be answered regarding the causation for the prevented widespread use of IOCS in daily practice worldwide: Does IOCS in cancer surgery affect survival, metastasis, and cancer recurrence rates? Does a reduction of allogeneic transfusion that might be achieved by IOCS in cancer surgery lead to better oncological outcome primarily via a reduction of potential adverse effects of allogeneic transfusion? i.e., what is safer for the patient: allogeneic or IOCS blood?

To date, the outcomes of patients treated with or without IOCS are unknown. Multicenter randomized double-blinded controlled trials are needed. To assess the effect of IOCS in colorectal cancer surgery, assuming a 5-year survival rate of 40%, at least 1,000 cases would be required to detect and demonstrate a presumed 10% change in the survival rate [13, 14]. For a mixed tumor collective with greater survival rates such as in prostate cancer, a greater number would be required to improve statistical power. Our previous 2020 meta-analysis of 27 observational studies and 1,606 IOCS treatments across a number of surgical specialties suggested that cancer recurrence and survival after nonmodified IOCS are not inferior compared to intraoperative allogeneic RBC transfusion, no transfusion, or preoperative autologous donation (PAD) [15]. On the opposite, we found a slightly reduced recurrence rate in this small number of observational cohorts. This new analysis includes additional observational studies and new IOCS-treated patients published since the prior meta-analysis to confirm or correct the previous result.

The current literature since our last meta-analysis (May 2019 to March 2022) was systematically reviewed by two independent authors. All full-text publications in PubMed, Cochrane, Cochrane Reviews, and Web of Science with the primary keywords “autologous transfusion/autotransfusion/cell salvage” and the links “tumor, cancer, metastasis, outcome, oncology, recurrence, survival” were screened. Data extraction was done from full-text eligible publications in English, French, Spanish, Italian, and German in a Web-embedded database. The following study details and population demographic characteristics were extracted: study design, timeline, sample size, number in each intervention arm, kind of control group, median length of follow-up in the IOCS group, groupwise absolute numbers for cancer recurrence, mortality or survival, allogeneic-transfused patients’ means and standard deviations for transfusion requirements (volume in mL), and length of hospital stay (LOS).

The quality of the evidence was rated as “high,” “moderate,” “low,” “very low” by both authors in the screening process according to a modified Grading of Recommendations Assessment, Development and Evaluation (GRADE) system. The risk of bias was not included in the assessment because the studies were exclusively observational studies, and therefore, we assumed a high risk of bias.

All studies with retransfusion of the salvaged blood with or without LDF filtration in tumor surgery were included in a systematic search, in which the study objective was survival/mortality, recurrence rate, transfusion requirements, and LOS. Studies were excluded if they were non-full-text articles, abstracts, poster presentations, and mixed surgeries for both benign and malign diseases. However, previously published meta-analyses, abstracts only, and case reports were included for qualitative analysis and were not included in the quantitative analysis. The effect estimates of the individual studies were summarized in a meta-analysis.

In some analyses, a distinction was made in the IOCS group between patients who were transfused with autologous blood and those who were not, and an “intention-to-treat” analysis was conducted, regardless of the actual “per-protocol treatment” provided (e.g., [16]). The same was applied in case of clinically irrelevant transfusion volumes and/or rates in the group of IOCS (e.g., [17], in whom only 5 out of 16 patients were retransfused with salvaged blood). If locoregional metastases and distant metastases were reported as outcome parameters, the rate of distant metastases was used. Overall mortality was calculated from the overall survival rate, if it was not stated separately. If general survival or mortality data and disease-specific data were reported, the general ones were used. For case-control studies, the matched results were used (e.g., [18]).

With regards to allogeneic transfusion requirements, some publications provided information on the transfusion volume and others on the rate of allogeneic transfusion recipients. Blood components such as RBC, platelet, and fresh frozen plasma were summed and estimated at 275 mL and whole blood at 450 mL, in order to establish comparability of the total transfusion volume. If a distinction was made between intraoperative and postoperative transfusion requirements, the intraoperative data were used, otherwise, we assumed the total transfusion requirement until discharge or the end of the observation period. If the transfusion rate was zero in a study, a continuity correction of 0.5 was conducted to calculate a risk estimate.

For binary endpoints (recurrences, mortality, patients requiring transfusion), the odds ratio (OR) with 95% confidence intervals (CI) was used. The mean difference (MD) with 95% CIs was used to calculate continuous data such as volume and LOS. The OR and Mantel-Haenszel method were also used to calculate blood volume and LOS. An estimation for MD was calculated using the inverse-variance method. All meta-analyses were based on a random-effects model. Statistical heterogeneity between studies was examined using the I², with the following overlapping categorization: I² of 0–30% characterizes little to no heterogeneity, 30–60% for moderate statistical heterogeneity, 50–90% for substantial statistical heterogeneity, and greater than 75% for significant statistical heterogeneity.

The updated meta-analysis included 34 observational studies, with 8,503 enrolled subjects, and 3,161 of those patients were IOCS-treated patients (see Table 1, search methodology Fig. 1). From all studies, 27 studies with n = 2,181 IOCS-treated subjects reported recurrence, 22 studies with n = 1,610 in the IOCS group reported mortality, 20 studies reported the number of transfused patients (n = 1,973 IOCS), 10 studies with n = 842 IOCS subjects reported transfusion volumes, and 10 studies (n = 401 IOCS) stated LOS.

In comparison to the last meta-analysis published in 2020 [15], 9 full-text articles (observational studies) with additional n = 653 IOCS-treated subjects were found. Of those, 7 articles were published after the last analysis, while two studies were included due to a wider inclusion criterion. A most recent study with 127 matched pairs could not be included for the recurrence and mortality analysis since the study estimated “time-to-event” data from a model [19].

We included two study pairs from the same institution (Kwon et al. [20] and Han et al. [21], Fujimoto et al. [18] and Hirano et al. [56]) despite a potentially overlapping patient population (overlap by 74 vs. 222, 46 vs. 50 IOCS-treated and analyzed subjects). Since enrollment periods and numbers of enrolled subjects varied, we assumed two different populations. All studies included observational data of varying cancer surgery for a wide range of cancer types. We also included studies about procedures associated with higher blood loss such as liver transplantation or spine surgery. The median observation period after surgery was almost 3 years: 35.6 months (1., 3. quartiles 23–48 months).

The results of the risk estimation are presented in Figures 2-6, a series of forest plots generated from included cancer surgeries for various reported outcomes comparing IOCS use and retransfusion versus control (no transfusion, allogeneic transfusion only), presented as ORs and 95% CI. The cancer recurrence rate decreased in the IOCS groups (OR 0.76 [95% [CI]: 0.64–0.94], heterogeneity I² = 0%). There was no observed difference between the groups regarding mortality (OR 0.95, 95% CI: 0.71–1.27) and LOS (MD −0.07, 95% CI: −0.63 to 0.48). Due to the high amount of heterogeneity between the studies, we did not calculate a pooled effect for transfusion rate (I² = 87%) and transfused volume (I² = 100%).

This updated meta-analysis on IOCS use for cancer surgery identified 7 new studies and 520 more IOCS-treated cancer patients (to a total sum of n = 3,161). Combining the outcomes from 34 studies, the result of the previous meta-analyses could be confirmed: IOCS in cancer surgery is associated with a reduced risk ratio for cancer recurrence and metastasis [15, 57]. Mortality and cancer-free survival episodes are not influenced by the use or avoidance of IOCS. The homogeneity of results from available trials suggests that tumor cells in IOCS blood may not impact survival and cancer recurrence. The most recent study by Nutu et al. [19] on IOCS use for hepatocellular cancer is consistent with our results [19]. However, these findings are based on a low level of evidence since quality of data is derived from observational studies. The only study protocol of RCT for IOCS in ovarian cancer (entitled “TICTOC,” n = 30 vs. n = 30 controls) that can be found is a feasibility trial not yet published [58]. Although it is important to note that there could be confounding results since in almost a third of the observational studies, nontransfused patients were compared to IOCS-treated subjects. Nevertheless, our meta-analysis is based on the best evidence available as randomized trials have yet to be done.

IOCS was used in many institutions exclusively on high blood loss situations (e.g., [22]) and, in most instances, accompanied or augmented with allogeneic blood transfusion. In a subgroup with comparable blood loss in the control and IOCS patients, the usage of IOCS tends to reduce allogeneic transfusion rate. This aspect was already shown for well-standardized surgical procedures such as liver transplantation for malignancy. However, in mixed data such as in this analysis, trials dealing with transfusion rates and blood requirement were too heterogeneous to draw conclusions. Surprisingly, this meta-analysis did not identify allogeneic transfusion as a risk factor for an increased recurrence rate, which stands in contrast to most studies on transfusion effects in oncologic surgery [1-6]. Since a large number of studies of this meta-analysis compared patients treated with IOCS supplemented with allogeneic transfusion to groups of patients not transfused, allogeneic transfused only, or transfused with predonated autologous blood, this design is not appropriate to assess allogeneic transfusion risks. As in many meta-analysis, the study protocols have to be considered heterogeneous: Included are control groups without any transfusion or the selection of IOCS patients by blood loss extend, coming down to an unfavorable selection bias according to the major surgical trauma and/or more advanced malignancy state in the IOCS group. The majority of retrospective studies identified the use or nonuse of IOCS and presented the institution’s standards as the reason for group allocation. It has to be considered that a majority of “controls” are patients bled sparingly intraoperatively; thus, IOCS was not used. However, there was a reduced recurrence rate of patients receiving autologous IOCS blood. The immune modulation by allogeneic blood would be the most plausible explanation for the group difference to the subgroups of controls of transfused subjects. The methodological description of observational studies rarely quantified the extent of allogeneic augmentation following autologous predonation. Patients without transfusion were too few (for recurrence 3.8%) to impact the overall result.

The concern of tumor dissemination or plantation of distant metastases by reinfusion of IOCS blood is based on physiological assumptions and is not well investigated. The warnings against the use of IOCS in cancer surgery are based on a single published case report that was cited when warnings against IOCS use are issued. On closer inspection, there is a misinterpretation of the results in the case report by Yaw et al. from 1975 [59]. The case report describes the tragic death of a 52-year-old patient with lung cancer following a bilobectomy. Malignant cells were identified in a microscopic evaluation of the transfused blood. However, the collected blood was not retransfused to the patient and was discarded due to the diagnosis of the frozen section. Hence, the IOCS did not cause the fatal outcome.

In light of the favorable results in more than 2,500 cancer patients and the fact that no single case report of cancer spread by IOCS use exists, theoretical concerns about the spread of autologous malign cells by IOCS lose weight. Since randomized controlled trials are difficult to conduct in these settings, this analysis of observational trials is the best available approach despite low numbers and high heterogeneity in the included observations.

Due to the limitation in data quality, we conclude that the safest option is to use IOCS in combination with LDF filtration. The filtration efficacy in terms of cancer cell reduction using a newer generation of LDFs is 99.6% [60] to 99.9% [61]. This was demonstrated for a large variety of tissues such as prostate and renal cancer [62]; osteosarcoma [63]; breast [64]; colorectal [65]; pancreatic [23]; hepatocellular [66]; endometrium, cervical, and ovarian [67]; spinal metastases [68]; and lung cancer [69].

In conclusion, the use of IOCS in cancer surgery in many observational studies was shown to be safe and positive in terms of a favorable outcome. Nevertheless, data quality is low, and the theoretical risk to cause hematogenous spread of malignant cells by reinfusion of the IOCS product is to be considered. Hence, multicenter randomized controlled trials are urgently needed to confirm the results of this meta-analysis. Meanwhile, the recommendation of some societies and groups of experts is to use IOCS in cancer surgery with some restrictions [70-75].

We thank the German Interdisciplinary Task Force for Clinical Hemotherapy IAKH for funding and support.

An ethics statement is not applicable because this study is based exclusively on published literature.

T.F.: honoraria and reimbursements for travel expenses, lectures, investigator meetings, and presentations from Janssen-Cilag, AstraZeneca, Vifor Pharma, Pharmacosmos, the German Red Cross, Aspect Medical, Organon, Alliance Pharmaceuticals, and Baxter Healthcare Corp; expert consulting contracts with Janssen-Cilag, Vifor Pharma, Pharmacosmos; research grants from the Else-Groenert-Foundation and the University Medicine Mannheim, University of Heidelberg. A.U.S.: research grant from Pharmacosmos, Denmark, to perform a single-center, prospective trial on preoperative anemia treatment. A.U.S. is supported by the German Research Foundation (Deutsche Forschungsgemeinschaft) grant STE 1895/9-1 and STE 1895/10-1 as part of the DFG research consortium FerrOS-FOR5146. A.H.: no conflicts of interest to declare. M.M.: no conflicts of interest to declare. G.D.: no conflicts of interest to declare. M.A.W.: no conflicts of interest to declare. J.H.W.: no conflicts of interest to declare. D.F.: no conflicts of interest to declare.

The German Interdisciplinary Task Force for Clinical Hemotherapy IAKH supports the topic by covering for cost during research, analysis, writing, and publication such as publication fees, salaries, article prints, and presentation expenses.

T.F.: study director and manager, project design, communication with the IAKH, selection of reviewers and authors, research and data extraction, manuscript drafting, and final check. A.U.S.: manuscript and scientific content control. M.M.: statistics. A.H.: manuscript native language check. G.D.: project design, scientific content check, and manuscript. M.A.W.: scientific content check and manuscript refinement. J.H.W.: project design and manuscript comments check. D.F.: analysis and data check, research and data extraction, manuscript drafting and refinement, and author coordination.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.