Study Type – Therapy (individual cohort)
Level of Evidence 2b
Study Type – Therapy (individual cohort)
To determine whether robotic-assisted laparoscopic radical prostatectomy (RALP) is associated with a lower transfusion rate than radical retropubic prostatectomy (RRP).
PATIENTS AND METHODS
In this cohort study, we evaluated 1244 consecutive patients who underwent RALP (830) or RRP (414) between June 2003 and July 2006. Demographics, clinical characteristics, pathology, blood loss and transfusion data were collected prospectively. Groups were compared for baseline characteristics, blood loss, change in haematocrit and transfusion using univariate statistics, and an exploratory multivariate model was developed.
RALP was associated with lower blood loss (median 100 vs 450 mL, P < 0.001) and a smaller change in haematocrit (median 7% vs 10%, P < 0.001) than RRP. Although both groups had low transfusion rates, the RALP group required fewer transfusions than the RRP group (0.8% vs 3.4%, P= 0.002). On univariate analysis, surgical approach (RRP vs RALP), estimated blood loss ≥500 mL and change in haematocrit ≥10% were the only the significant predictors of transfusion. In the exploratory multivariate model RALP was the only significant predictor of reduced need for transfusion, with an odds ratio of 0.23 (95% confidence interval 0.09–0.58; P= 0.002).
This study shows that RALP is associated not only with less blood loss and a smaller decrease in haematocrit, but also a decreased need for transfusion.
(retropubic) radical prostatectomy
robotic-assisted laparoscopic RP
estimated blood loss
body mass index
seminal vesicle involvement
lymph node involvement
Radical prostatectomy (RP) has been the standard surgical treatment for clinically localized prostate cancer for several decades. The increase in PSA-based screening together with a reduction in the threshold of indications for prostate biopsy has led to an increase in diagnosis, as well as migration to earlier stages of disease at the time of diagnosis. Consequently, the number of candidates for RP has also increased . This, in turn, has led to the search for reducing the invasiveness of open surgery and improving functional results. With these intentions, first laparoscopic then robotic systems were introduced and gained wide acceptance . Despite the recent advances in operative technique, blood loss remains the most common intraoperative complication and sometimes warrants blood transfusion .
Many strategies have been suggested to decrease blood loss, including erythropoietin, controlled hypotension and acute normovolaemic haemodilution, autologous donation, and the cell-saver autotransfusion . Many studies have attempted to predict the risk factors for blood loss to decrease the morbidity associated with RP and to take more cost-efficient measures. However, risk factors for predicting blood loss remain elusive [4,5]. The safety of homologous transfusion has improved in recent years, but the possibility of having transfusion-related reactions or acquiring transfusion-transmitted diseases, together with the uncertainty of whether blood products will be required during surgery, still bother patients and can cause substantial anxiety [3,5,6].
Robotic-assisted laparoscopic RP (RALP) has been studied extensively as an alternative to open retropubic RP (RRP). RALP is associated with decreased blood loss in several cohort studies, but there are fewer data available comparing transfusion rates between RALP and RRP, with no randomized controlled studies comparing the procedures [2,7]. Therefore, we conducted the present prospective cohort study to determine whether there is a difference in transfusion rate between RALP and RRP in a large series from a referral centre with a high volume of both RRP and RALP. In conducting this study, we sought to identify possible patient, disease and surgical characteristics associated with transfusion requirement.
PATIENTS AND METHODS
Between June 2003 and July 2006, 830 consecutive men underwent RALP and 414 RRP with lymphadenectomy for clinically localized prostate cancer at Vanderbilt University Medical Center. Clinical, demographic, perioperative and pathological data were collected prospectively into an database approved by the institutional review board. The surgical approach was selected by the patient after a discussion of the risks and benefits of each alternative. During this period, one surgeon performed only RALP (S.D.H.), two only RRP (S.S.C., M.S.C.) and one performed both procedures (J.A.S.). All patients received general anaesthesia and no epidural catheters were used either during or after RP. A closed suction drain was placed at the time of surgery and removed before discharge in all patients. Estimated blood loss (EBL) and operative duration were recorded as documented by the anaesthesiologist. The serum haematocrit was obtained before RP and the morning afterward; the change in haematocrit was calculated as the difference before and after RP. Although there was no pre-specified criterion for transfusion, a postoperative serum haematocrit of <28% was generally considered an indication for transfusion. Ultimately, transfusion was administered at the discretion of the surgeon.
Patient characteristics and pathological variables were compared across treatment groups using the Kruskal–Wallis test for continuous variables and Fisher’s exact test for categorical variables. Associations between risk of transfusion and clinical, surgical and pathological variables, i.e. RALP vs RRP, patient age, body mass index (BMI), year of surgery, surgeon, PSA level, clinically palpable disease, neoadjuvant hormonal therapy, previous radiotherapy, biopsy Gleason score ≥7, prostate volume, tumour volume, pathological Gleason score ≥7, extracapsular extension, (ECE), seminal vesicle involvement (SVI), were tested with univariate logistic regression models and Fisher’s exact test. We also evaluated the association between EBL and risk of transfusion by dichotomizing both EBL and change in haematocrit, using ≤500 vs >500 mL and ≤10% vs >10%, respectively. As the number of transfusions (21 events) was quite low only an exploratory multivariate model could be constructed. To ameliorate over-fitting, we used a minimum number of variables in the model, including RALP vs RRP, age, PSA level, pathological Gleason score ≥7 and ECE.
The two groups were similar in age, race and BMI (Table 1), but the RRP group had a significantly higher median baseline PSA level, a higher proportion with clinically palpable disease, and more aggressive pathological features, including higher pathological Gleason score, and higher proportion of patients with ECE, SVI, lymph node involvement (LNI) and positive surgical margins (Table 1). The median prostate volume was significantly larger in the RALP group, with a median (interquartile range, IQR) of 46 (37–58) vs 41 (31–52) mL, respectively (P < 0.001). Twenty-five patients (3.7%) in the RALP group and 20 (6.0%) in the RRP group had undergone neoadjuvant hormonal therapy or preoperative radiotherapy (Table 1).
|Mean (sd), median (IQR) or n (%) variable||RRP||RALP||P|
|No. of patients||414||830|
|Age, years||61.5 (7.5)||60.5 (7.2)||0.100|
|Non-White||43 (10.4)||50 (6.0)||0.008|
|BMI, kg/m2||28.0 (4.6)||28.2 (4.2)||0.272|
|PSA, ng/mL||6.0 (4.6–9.1)||5.5 (4.4–7.3)||<0.001|
|Clinically palpable (≥cT2)||128 (31.2)||204 (24.8)||0.017|
|Biopsy Gleason score|
|≤6||261 (63.0)||578 (69.8)||<0.001|
|7||104 (25.1)||211 (25.5)|
|8–10||49 (11.8)||39 (4.7)|
|Neoadjuvant hormonal therapy||20 (4.8)||29 (3.5)||0.279|
|Preoperative radiotherapy||5 (1.2)||2 (0.2)||0.044|
|Prostate volume, mL||41 (31–52)||46 (37–58)||<0.001|
|≤6||186 (45.3)||450 (54.7)||<0.001|
|7||167 (40.6)||312 (37.9)|
|8–10||58 (14.1)||61 (7.4)|
|ECE||117 (28.3)||155 (18.7)||<0.001|
|SVI||48 (11.6)||35 (4.2)||<0.001|
|LNI||17 (4.1)||0 (0)||<0.001|
|Positive margin||132 (31.2)||171 (20.6)||<0.001|
|EBL, mL||450 (300–600)||100 (50–200)||<0.001|
|Change in haematocrit, %||10 (8–12)||7 (6–9.5)||<0.001|
|Transfusion||14 (3.4)||7 (0.8)||0.002|
RALP was associated with a significantly lower EBL (median 100 vs 450 mL, P < 0.001) and change in haematocrit (median decrease 7% vs 10%, P < 0.001) than RRP (Table 1). In all, 21 patients had a blood transfusion (1.7%). Although both groups had low transfusion rates, the RALP group required fewer transfusions than the RRP group (0.8% vs 3.4%, P= 0.002; Table 1).
We then sought univariate associations with transfusion (Table 2); transfusion was associated with procedure type (odds ratio 0.24, 95% CI 0.10–0.61; P= 0.002 for RALP vs RRP), EBL (7.4, 3.0–18.0, P < 0.001 for EBL >500 vs ≤500 mL) and decrease in haematocrit (17.0, 4.0–73.6, P < 0.001 for >10% vs ≤10%). No other baseline or pathological characteristics were significantly associated with transfusion. Only one of 49 patients who had neoadjuvant hormonal therapy received a transfusion and none of the patients who had preoperative radiotherapy received a transfusion. There were no positive lymph nodes among RALP patients and no transfusions among the 17 patients with positive lymph nodes in the RRP group. The likelihood of transfusions was not associated with surgeon for the group as a whole (Fisher’s exact P= 0.164), for RRP patients (P > 0.99) or for RALP patients (P= 0.610). There was a trend of decreasing likelihood of transfusion with advancing calendar year (Mantel-Haenszel P < 0.001) because of increasing number of RALPs performed in the latter years (47% in 2003, to 77% in 2006).
|Variable||Odds ratio (95% CI)||P|
|RALP vs RRP||0.24 (0.10–0.61)||0.002|
|Age (continuous)||1.04 (0.98–1.11)||0.146|
|BMI (continuous)||0.97 (0.86–1.10)||0.585|
|Date of surgery (continuous)||1.00 (0.99–1.00)||0.744|
|PSA (continuous)||0.99 (0.92–1.07)||0.873|
|Clinically palpable||1.37 (0.55–3.41)||0.504|
|Biopsy Gleason ≥7||0.83 (0.32–2.15)||0.699|
|Prostate volume (continuous)||0.98 (0.96–1.01)||0.231|
|Tumour volume (continuous)||1.03 (0.98–1.08)||0.295|
|Pathological Gleason score ≥7||0.79 (0.33–1.90)||0.605|
|Positive margin||0.97 (0.35–2.67)||0.953|
|EBL >500 mL||7.40 (3.03–18.0)||<0.001|
|Change in haematocrit >10%||17.0 (3.95–73.6)||<0.001|
|RALP vs RRP||0.23 (0.09–0.58)||0.002|
|Age (continuous)||1.04 (0.98–1.11)||0.220|
|PSA (continuous)||0.99 (0.91–1.08)||0.825|
|Pathological Gleason score ≥7||0.69 (0.27–1.80)||0.447|
Because there were only 21 events (transfusions) in this cohort, there was a limited possibility of a multivariate analysis with more than two covariates to identify predictors of transfusion. Nonetheless, we fitted an exploratory multivariate model, which included age, PSA level, pathological Gleason score, ECE and procedure type. Only procedure type was significant in this model (odds ratio 0.23, 95% CI 0.09–0.58, P= 0.002 for RALP vs RRP). We re-ran the model, including only procedure plus one additional variable at a time; this did not alter the fact that RALP was the only significant predictor of reduced need for transfusion (Table 2).
Clear visualization during any surgery is necessary to obtain both optimal oncological and functional results. Intraoperative bleeding can obscure the operative field, leading to increased transfusion requirements, and both peri- and postoperative morbidity . Bleeding during RP is highly variable and usually arises from inadvertent vascular injury to venous structures. It can also occur during lymphadenectomy if the external iliac vein or branches of the hypogastric artery are damaged . Several studies have shown that blood loss during RP is influenced by many factors, including surgical experience, surgical approach (intra- vs extraperitoneal), neurovascular bundle preservation, use of neoadjuvant hormonal therapy, use of general anaesthesia, prostate size, BMI, a marked prominence of apical periprostatic veins on preoperative endorectal MRI, and operative duration [3,4,8–11]. However, perioperative bleeding has been noticeably reduced by better appreciation of the anatomy of the dorsal venous complex, apex and neurovascular bundles [2,12,13]. Magnified vision, the positive pressure of the pneumoperitoneum, new haemostatic energy sources and materials, and head-down position of the patient obtained during minimally invasive surgery also help to gain better haemostasis .
In a meta-analysis of recent RP series, Ficarra et al. reported a median EBL of 385–1550 mL in RRP, 189–1100 mL in laparoscopic RP and 103–609 mL in RALP. RALP patients had significantly less blood loss than RRP patients. However, cumulative analyses were not possible because of insufficient data in a proper format in the included studies. Similarly, Parsons and Bennett  reported a similar meta-analysis and concluded that RALP was associated with significantly less operative blood loss than RRP (standardized mean difference −1.58, CI −2.07 to −1.08, P < 0.001). Our data corroborate these findings; the median (IQR) EBL was 100 (50–200) mL in the RALP group and 450 (300–600) mL in the RRP group (P < 0.001).
Institutions use different methods to for the EBL during surgery and there is no standard. Brecher et al. reported that EBL recorded by anaesthesiologists correlates with the true blood loss. Haematocrit levels before and after RP were also included in our study, to provide more reliable results for EBL and transfusion requirements. The change in haematocrit was significantly lower in the RALP than RRP group in the present study (median 7% vs 10%, P < 0.001), strengthening our results for EBL.
The threshold for transfusion varies across institutions and should be personally adapted to patient needs and surgeons’ discretion. Hebert et al. randomized critically ill patients to a blood transfusion policy either that was restrictive (haemoglobin <7 g/dL) or liberal (haemoglobin <10 g/dL). They reported that the restrictive blood transfusion strategy is safe and possibly better than the more liberal approach. While a haemodynamically stable patient requires a blood transfusion when the haemoglobin level is <7 g/dL, a patient with ischaemic heart disease needs it at a haemoglobin level below 8–9 g/dL. Hogue et al. reported that during and after RRP, a haematocrit of <28% was independently associated with a higher risk fo myocardial ischaemia. Thus, transfusion requirement reflects the impact of the intraoperative blood loss . McClinton et al. reported perioperative blood transfusion to be associated with adverse effects on the survival of patients with prostate cancer undergoing surgery. While Oefelein et al. also reported decreased survival with increasing blood loss (risk ratio 1.08, 95% CI 1.05–1.10, for every 100 mL) they found that intraoperative transfusion was not associated with a higher likelihood of recurrence in a model that controlled for blood loss (in fact, the risk of recurrence decreased significantly with each unit transfused, whether the blood was autologous or allogenic). They concluded that adverse outcomes might be related to more extensive disease and tumour characteristics rather than immunological effects of transfusion. Together with these concerns, the cost, risk of transfusion reaction and patient anxiety about acquiring blood-borne diseases, make transfusion a highly relevant topic.
Numerous strategies have been proposed to minimize the transfusion requirements, such as acute normovolaemic haemodilution, preoperative autologous donation, intraoperative cell salvage with autotransfusion, and preoperative erythropoietin therapy [4,8,9]. While each of these strategies has its merits, none of these techniques has proven sufficiently cost-effective and effective to gain wide acceptance. In any event, the likelihood of a blood transfusion with RP has decreased over time from 62–89% in the late 1980s to 1–3.5% in late 1990s . Ficarra et al., in their meta-analysis, reported blood transfusion rates of 9–29% for RRP, 1–5% for LRP, and 0–2.6% for RALP. They also reported transfusion rates significantly higher in RRP than RALP (relative risk 4.51, 95% CI 1.35–15.03; P= 0.01). However, sensitivity analysis limited to prospective studies shown only a statistically insignificant difference in favour of RALP (relative risk 7.68, 95% CI 0.62–95.1; P= 0.11). Parsons et al., in a similar analysis with 2869 patients, reported a 77% lower risk (relative risk 0.23, 95% CI 0.11–0.49; P < 0.001) and significantly lower incidence (relative difference −0.19, 95% CI −0.33 to −0.05; P= 0.008) in the RALP than RRP group. Sequential omission of each study from the analysis did not affect the results. Our results parallel these previous studies in identifying a transfusion rate of 0.8% for RALP and 3.4% for RRP (P < 0.002).
A few investigators have made an effort to identify predictors of perioperative transfusion in men undergoing RP. Dash et al. analysed prospectively collected data of 1123 consecutive RRP cases, reporting a 9.3% overall blood transfusion rate in their series. They found prostate size, surgeon experience, use of general anaesthesia and use of neoadjuvant hormonal therapy to be independently associated with homologous transfusion requirement. In a study of 436 consecutive RRP patients, Chang et al. reported that BMI was a significant predictor of EBL. In the present study, we found that surgical approach (RRP vs RALP), EBL >500 mL and change in haematocrit of >10% were the only factors associated with transfusion. The multivariate analysis was limited by the few events, but we found no other significant associations on univariate tests. Specifically, we did not find differences in transfusion requirement based on age, BMI, previous hormonal or radiotherapy, or disease characteristics.
Our study has several other limitations. Although it was a prospective study it was not randomized, and thus the groups might have differed in important baseline characteristics. However, we found no baseline or disease characteristics that were associated with the likelihood of transfusion, so this limitation might not have affected our results. The few transfusions in the cohort precluded construction of a formal multivariate model, so we were limited to an exploratory model.
In conclusion, this study showed that RALP is associated not only with less blood loss and a smaller decrease in haematocrit, but also a decreased need for transfusion. On univariate analysis, surgical approach (RRP vs RALP), EBL >500 mL and change in haematocrit of >10% were the only the factors associated with transfusion. In an exploratory multivariate model RALP was the only significant predictor of a reduced need for transfusion. Thus, further studies with more events are needed to identify the independent predictors of transfusion.
CONFLICT OF INTEREST