A retrospective evaluation of preoperative anemia in patients undergoing radical cystectomy for muscle-invasive urothelial urinary bladder cancer, with or without neoadjuvant chemotherapy

Background and objective Neoadjuvant chemotherapy (NAC) can be associated with anemia, which can lead to more perioperative blood transfusions (PBT). Usage of PBT is associated with worse oncological outcomes. We evaluated the prevalence of preoperative anemia (PA) and the effect on hemoglobin levels depending on surgery timing after NAC. Methods A retrospective single-center study with 240 consecutive patients undergoing radical cystectomy (RC) between 2001 and 2014 for muscle-invasive urothelial carcinoma (MIBC). Anemia was defined according to the WHO classification (male ≤ 130 g/L, female ≤ 120 g/L). Multivariable logistical regression was used to identify factors associated with PA and Pearson correlation for evaluating the change in hemoglobin levels depending on surgery timing. Results Overall, 128 (53.3 %) patients were anemic pre-RC and 87 (36.3 %) patients received NAC. In a multivariable analysis, age, receipt of NAC, female gender, and low BMI were independent predictors of PA. In patients receiving NAC, the time to surgery from the last NAC cycle was correlated with the change in hemoglobin levels between the initiation of NAC and surgery. Conclusions PA was common in patients undergoing RC for MIBC. Receipt of NAC was found to be a strong predictor of PA. Clinical message The emerging treatment of cisplatin based neoadjuvant chemotherapy for muscle-invasive bladder cancer, confers an increased risk for preoperative anemia. In the management of this malignancy, preoperative anemia renders further attention and focus.


Background
Urothelial bladder cancer (UBC) is the fourth most common cancer in the Western World. The majority of patients diagnosed have non-invasive UBC, but around 25 % are muscle-invasive (MIBC) at time of diagnosis (Burger et al. 2013). The recommended treatment for patients with MIBC is radical cystectomy (RC) with lymph node dissection within 3 months to avoid progression and to decrease mortality. Addition of neoadjuvant chemotherapy (NAC) with cisplatin-based combinations is recommended for patients fit for chemotherapy (Witjes et al. 2015). This recommendation is based on the results of prospective randomized trials showing an absolute increase of survival with 5-8 %, 5 years median observation time (Sherif et al. 2004;Vale 2005). RC is considered major surgery and can be associated with significant blood loss and need for preoperative blood transfusions (PBT) (Shabsigh et al. 2009). The use of PBT have been shown to have a negative effect on prognosis in UBC (Abel et al. 2014;Linder et al. 2013;Gierth et al. 2014;Morgan et al. 2013). This detrimental effect has previously also been shown in other cancer types; gastric, colorectal and hepatocellular carcinoma (Ojima et al. 2009;Amato and Pescatori 2006;Wang et al. 2009). The mechanism of this unfavorable outcome is not fully understood. Negative immunomodulatory effects following transfusion have been suggested as one reason (Vamvakas and Blajchman 2007). Difficult surgical conditions in more advanced tumors or in patients with a history of abdominal surgery or radiation therapy of the pelvic organs, may also be a reason for increased use of PBT. Thus adding confounding factors to the long term results following PBT. Preoperative anemia (PA) has also been suggested to be an independent predictor for the outcome following RC. A recent study showed that PA is associated with worse oncological outcomes in patients undergoing RC (Gierth et al. 2015). In order to improve treatment in urothelial MIBC, it is valuable to have patients optimized prior to RC. The use of NAC has postulated side effects such as anemia and this, in combination with a risk of significant blood loss during surgery, can increase the risk for PBT. We sought to evaluate the prevalence of PA, identify associated factors and evaluate the effect of NAC on hemoglobin (Hb) levels in patients with urothelial MIBC undergoing RC.

Methods
In this retrospective single-center study we analyzed all RCs due to MIBC from 2001 to 2014. From the database we identified 349 consecutive patients, undergoing RC at a tertiary referral center (Norrland university hospital in Umeå, Sweden). We excluded 34 patients with nonurothelial cancer, 71 patients with non-invasive UBC and 4 patients who underwent salvage cystectomy. In summary, 240 consecutive patients constitute the study cohort. The study included 176 (73.3 %) male and 64 (26.7 %) female patients and the median age was 69 years (IQR 62-75). Clinicopathological variables were gathered from patient journals regarding age, gender, BMI (Body Mass Index), ASA (American Society of Anesthesiologists) score, comorbidities (previous heart attack, hypertonia, atrial fibrillation, asthma/Chronic obstructive pulmonary disease (COPD), Diabetes Mellitus), hydronephrosis, estimated blood loss (EBL), receipt of PBT and number of units, receipt of NAC, clinical tumor stage (cT), Hb prior to first NAC-cycle, preoperative Hb and year of surgery.
Information about the change in Hb levels from the initiation of NAC until RC and also time to surgery from last NAC-cycle, was also evaluated. Hydronephrosis was defined as history of hydronephrosis prior to RC. PBT was defined as transfusion of allogenic red blood cells during the operation or in the postoperative hospitalization. The decision to administer blood transfusion was made by treating physicians. No standardized thresholds for transfusion were in place during the study period.

Statistical analyses
Patients were categorized based on their Hb-level into two groups: Normal Hb-level or anemic. Anemia was defined as Hb ≤ 130 g/L in male and Hb ≤ 120 g/L in female patients, according to the WHO classification (Blanc et al. 1968). Clinical and pathologic characteristics were compared between the groups using Mann-Whitney U test for continuous variables and Fisher's exact test for categorical variables. Spearman's rank correlation was used to check the development of EBL and usage of units of PBT over time. Univariable and multivariable logistic regression analysis was performed to identify factors associated with PA. In NAC-patients, Pearson correlation was used to analyze the relationship between the change in Hb-levels following NAC and time to RC. The reported p values were two-sided and statistical significance was set at 0.05. All statistical analyses were made using SPSS Statistics ® 22 (SPSS, IBM Corp, Armonk, NY, USA).

Development of NAC, anemia, PBT and EBL over time
The use of NAC increased during 2001-2014. The first patient receiving NAC was in 2003, with a steady increase during the years. In 2014, 81.3 % of patients with MIBC received NAC at this center (Fig. 1a). There was no statistically significant increase in the prevalence of PA over time (p = 0.067; Fig. 1b). Overall, the median EBL was 1000 ml (IQR 700-1712.5). There was no difference in EBL between patients receiving NAC compared to the non-NAC group (p = 0.369). The median EBL decreased from 2500 (IQR 2000-11,500) Table 3 shows the details of NAC treatment. In NACpatients (n = 87), the median time to surgery from the last cycle of NAC was 35 days (IQR 29-42). The median change in Hb-level was -15 g/L (IQR −26 to −3.5). The change in Hb-levels can also be expressed in that the number of patients with anemia prior to NAC was 48 (55.2 %), compared to 70 (80.5 %) who had anemia prior to RC post-NAC. The change in Hb following NAC was correlated with time to RC (Pearson correlation coefficient: 0.221, p value: 0.042, Fig. 2a). There was an upward trend towards an increasing Hb after 7 weeks (Fig. 2b).

Discussion
The increased risk of adverse outcomes following PBT in conjunction with RC for MIBC, needs to be addressed. Utilization of PBT affects the long term results, not only pertaining to overall and cancer-specific mortality but also to the long term risk of tumor recurrence (7). In general, PA increases the risk for usage of PBT, and we found it of importance to investigate effects of NAC on PA. This with background of optimizing this rapidly emerging treatment, which ultimately aims to improve both overall and cancer-specific mortality. We found that 53.3 % of all the patients had PA. This is a higher prevalence than found in other studies, having reported rates ranging from 39.3 to 51.5 % (Gierth et al. 2014(Gierth et al. , 2015Moschini et al. 2015). However, two of the studies excluded patients receiving NAC and in the third study only 2.9 % received NAC. We identified four independent predictors of PA: Age, gender, receipt of NAC and BMI. Receipt of NAC was the strongest predictor (OR 9.668). Anemia has previously been shown to be more common with increasing age (Gierth et al. 2015). A lower BMI could reflect the patient´s state of health, which in turn can have an effect on Hb-levels. The fact that females had a lower risk of PA could partly be explained by a lower anemia threshold in the WHO-definition. Our current praxis is that RC is planned 4 weeks after final NAC-cycle. The planned time to RC in other centers in Sweden varies; some centers proceed after 6-8 weeks. The current EAUrecommendation regarding timing of RC is that the operation is not to be delayed more than 3 months. However, this recommendation is based on single-center studies on chemo naïve patients. A recent study explored the effect of delaying RC > 3 months on survival in a nationwide cohort (Bruins et al. 2016). The investigators also included NAC-patients, showing that the overall survival was similar between patients undergoing RC < 3 months compared to patients with RC > 3 months. This suggests that the 3-month recommendation may not be applicable for NAC-patients. On the other hand, in NAC-nonresponders, an increased delay to RC might have a negative effect on survival. The problem with non-responders is currently being handled differently, depending on local traditions. Computerized tomography is commonly used in this center after the second NAC-cycle, for identifying non-responders and in case of non-response or progress, we proceed directly to RC avoiding NAC-cycle 3. Another suggested method is cystoscopy, with a recent study showing that cystoscopy-findings after NAC-cycle 2 are independent predictors of extravesical disease and pathologic downstaging (Mansour et al. 2015). In our study we have seen that utilization of NAC has increased in the last decade. Thus one can assume that with more patients receiving NAC, PA could become more prevalent. Another interesting approach would be to evaluate differences in PA, comparing complete responders to non-responders, thus postulating that micrometastic disease in the latter subgroup might be reflected in PA. Yet in this limited material there was no difference. Further we did not observe any differences in EBL or the amount of PBT received in the NAC group compared to the no-NAC-group. One reason might be that the NACpatients in our material were significantly younger than the chemo naïve patients (p = 0.049). The amount of PBT decreased totally during the studied period. This is probably due to fewer surgeons performing RC following an ongoing national centralization of RC, an increased awareness of both perioperative bleeding and EBL due to national registers (The Swedish bladder cancer register with increasing national coverage from the 2000s and the Swedish cystectomy register which started in 2011) and further the rapidly increasing use of ileal conduit (86.5 % for 2013 in Sweden) instead of other more time consuming urinary diversions. Within the scope of prospective trials addressing these matters, consideration could be given to identify and treat anemia prior to RC, specifically without allogenic blood transfusions. Another way to address the described problem could be in adjusting the time to surgery after NAC, in order to let the Hb-level recover, and thus reduce the amount of patients with PA and the number of PBT required. In a few patients (n = 12) waiting for surgery more than 7 weeks after final NAC cycle, there  was a trend towards a smaller Hb-decrease post-NAC, compared with patients having a shorter interval to RC having a larger decrease. More data is needed to conclusively say if an increased time to surgery could affect the change in Hb-levels. More studies are also required to assess if delayed surgery following NAC has an impact on survival. The limitations of this study include its retrospective design on a relatively small study population. No pathological review was made, and due to the long study period multiple surgeons were involved. Furthermore no data on GFR were available which could add additional information regarding selection of patients to NAC and the individual effect of NAC on renal function. More data from centers where the planned time to RC after NAC is longer could also add value to the analysis of the impact of delayed surgery on PA and the amount of PBT.

Conclusions
PA was common in patients undergoing RC for MIBC. Receipt of NAC was found to be a strong predictor of PA. This has not been described previously in any evaluations pertaining to NAC-treatment of urothelial MIBC. Strategies to identify and manage patients with PA should be developed to avoid PBT in order to optimize management and prognosis, yet both larger retrospective and also prospective evaluations relating to PA and NAC, need first to be undertaken.