2.1 Group, NAC, Stage

We included 49 patients and after removing 1 patient for incompletness of key data, the final dataset included 48 patients, of which 19 (39.6%) received neoadjuvant chemotherapy (NAC, 2 to 6 cycles, median=4). Patients stage ranged from IB to IVB with most common IIIB in NAC and IIB in controls (V=0.62, p=0.003, Mann-Whitney p<0.001). Stage is significantly correlated with the number of neoadjuvant chemotherapy cycles (R_Spearman=0.562, p<0.001).

NAC patients received chemotherapy doses ranging from 30 to 40 (median=40) mg/mp, lower than controls (20 to 150, median=40, p=0.027). NAC patients received a statistically similar median number of chemotherapy cycles (2, ranging from 1 to 4) as the controls (3, 1 to 5, p=0.400). NAC patients lost sugnificalnty more of their planned cycles compared to controls (1 to 4, median=3 vs. 0 to 4, median=1, p<0.001). This resulted in significantly less total chemotherapy received (expresed as a fraction of the total planned dose) in NAC patients, compared to controls (20% to 80%, median=40%, vs 20% to 100%, median=60%, p<0.001). The total chemotherapy received is significantly correlated with the number of neoadjuvant chemotherapy cycles (R_Spearman=-0.610, p<0.001).

Toxicity was measured by blood cell counts, hemoglobin and creatinin measurements and urinary and digestive toxicity.

2.2 Leucocytes

NAC patients showed insignifficant toxicity by leucocyte counts (median=2440, 850:6310 vs median=2510, 1080:5600, p=0.393) as well as grade of leucopenia (median=2, 0:4 vs median=2, 0:3, p=0.250).

Lowest leucocyte count is not significantly correlated with the number of neoadjuvant chemotherapy cycles (R_Spearman=-0.211, p=0.149), even less after adjuting for total chemotherapy received (R_Spearman=0.015, p=0.920), but showed an increased correlation after adjuting for stage (R_Spearman=-0.409, p=0.004).

On average, every additional neadjuvant chemotherapy cycle decreased the neutrophile count by 107.63 ±85.4 cells/mL (non significant). After adjusting for total chemotherapy received the average WBC count decreased by 26.13 ±103.4 cells/mL (non significant) but after adjusting for stage, the average neutrophile count decreased by 244.62* ±94.7 cells/mL. However, when adjusting for both at the same time, the average neutrophile count decreased by 174.24 ±113.5 cells/mL (non significant).

There is a high degree of similarity between the Lowest Neutrophile count and the Lowest Leucocyte count (R=0.934***, the same as Lin’s CCC on standardized data) which explains the similarity of the results involving these two measurements and suggests that leucocyte toxicity is primarly based on neutrophiles.

2.3 Neutrophiles

NAC patients showed insignifficant toxicity by neutrophiles counts (median=1590, 540:5190 vs median=1690, 610:3970, p=0.292) as well as grade of neutropenia (median=1, 0:3 vs median=1, 0:3, p=0.242).

Lowest Neutrophile count is not significantly correlated with the number of neoadjuvant chemotherapy cycles (R_Spearman=-0.235, p=0.107), even less after adjuting for total chemotherapy received (R_Spearman=0.057, p=0.704), but showed an increased correlation after adjuting for stage (R_Spearman=-0.423, p=0.003 ).

On average, every additional neadjuvant chemotherapy cycle decreased the neutrophile count by 95.20 ±73.31 cells/mL (non significant). After adjusting for total chemotherapy received the average neutrophile count increased by 12.22 ±86.21 cells/mL (non significant) but after adjusting for stage, the average neutrophile count decreased by 234.40** ±78.53 cells/mL. However, when adjusting for both at the same time, the average neutrophile count increased by 135.03 ±91.59 cells/mL (non significant).

2.4 Lymphocytes

NAC patients showed insignifficant toxicity by lymphocyte counts (median=400, 70:950 vs median=390, 150:660, p=0.987) as well as grade of lymphopenia (median=3, 2:4 vs median=3, 1:4, p=0.381).

Lowest lymphocyte count is not significantly correlated with the number of neoadjuvant chemotherapy cycles (R_Spearman=0.077, p=0.601), even after adjuting for total chemotherapy received (R_Spearman=-0.073, p=0.627) nor after adjuting for stage (R_Spearman=0.180, p=0.225).

On average, every additional neadjuvant chemotherapy cycle increased the lymphocyte count by 2.52 ±11.1 cells/mL (non significant). After adjusting for total chemotherapy received the average lymphocyte count decreased by 9.61 ±13.36 cells/mL (non significant) and after adjusting for stage, the average lymphocyte count increased by 11.25 ±13.05 cells/mL (non significant). When adjusting for both at the same time, the average lymphocyte count decreased by 1.08 ±15.51 cells/mL (non significant).

2.5 Platelets

NAC patients showed signifficantly more toxicity by platelet counts (median=108, 36:208 vs median=165, 56:462, p=0.017) but not by grade of thrombocytopenia (median=0, 0:3 vs median=0, 0:3, p=0.987).

Lowest platelet count is significantly correlated with the number of neoadjuvant chemotherapy cycles (R_Spearman=-0.423, p=0.003), even after adjuting for total chemotherapy received (R_Spearman=-0.249, p=0.091) and after adjuting for stage (R_Spearman=-0.419, p=0.003).

On average, every additional neadjuvant chemotherapy cycle decreased the platelet count by 14.11** ±5.1 cells/mL. After adjusting for total chemotherapy received the average platelet count decreased by 14.41* ±6.3 cells/mL and after adjusting for stage, the average platelet count decreased by 17.85** ±6.0 cells/mL. When adjusting for both at the same time, the average platelet count decreased by 18.79* ±7.3 cells/mL.

2.6 Hemoglobin

NAC patients showed signifficantly lower hemoglobin (median=10.2, 8.7:11.9 vs median=11.6, 9:13.4, p=0.003) as well as a higher grade of anemia (median=1, 0:2 vs median=1, 0:2, p=0.025).

Lowest hemoglobin level is significantly correlated with the number of neoadjuvant chemotherapy cycles (R_Spearman=-0.484, p<0.001), even after adjuting for total chemotherapy received (R_Spearman=-0.371, p=0.010) and after adjuting for stage (R_Spearman=-0.001, p=0.031).

On average, every additional neadjuvant chemotherapy cycle decreased the hemoglobin level by 0.29*** ±0.08 cells/mL. After adjusting for total chemotherapy received the average hemoglobin level decreased by 0.25* ±0.10 cells/mL and after adjusting for stage, the average hemoglobin level decreased by 0.22* ±0.09 cells/mL. When adjusting for both at the same time, the average hemoglobin level decreased by 0.17 ±0.11 cells/mL (p<0.1).

2.7 Other toxicities

In both groups, 10% of the patients had elevated creatinin levels. Among cases, 31.6% had registered urinary toxicity grade 1 (none with grade 2) while among controls 31% had toxicity grade 1 and another patient had grade 2 (p=0.787).

Digestive toxicity grade > 0 were found in 68.4% on the NAT patients (with 15.8% grade 2) and 55.2% (with 10.3% grade 2) of the control patients (p=348). RT was given to Lombo-aortics in 47.4% of the NAC patients and 10.3% of the control patinents (OR=7.8, p=0.006).

2.8 Notes:

A patient had no record of stage and was removed form the analisys. Another patient had radiotherapy duration of 400 days and was left blank.

3.1 NAC ~ Stage

## 
##  Spearman's rank correlation rho
## 
## data:  baza_de_date$`Neoadjuvant chemotherapy` and baza_de_date$Stage.numeric
## S = 8000, p-value = 0.00003
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##   rho 
## 0.562

## 
##  Wilcoxon rank sum test with continuity correction
## 
## data:  Stage.numeric by Group
## W = 100, p-value = 0.00007
## alternative hypothesis: true location shift is not equal to 0

Figure. 1: Correlation between Stage and the number of NaC cycles. Size reprezents number of patients. Boxplot by group. Control: no NAC cycles. In NAC cases, median and IQR are at the IIIB level.

3.2 NAC ~ Total chemotherapy received (fraction of planned total dose)

## 
##  Spearman's rank correlation rho
## 
## data:  baza_de_date$`Neoadjuvant chemotherapy` and baza_de_date$`Total chemotherapy received`
## S = 30000, p-value = 0.000004
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##   rho 
## -0.61

## 
##  Wilcoxon rank sum test with continuity correction
## 
## data:  Total chemotherapy received by Group
## W = 500, p-value = 0.00009
## alternative hypothesis: true location shift is not equal to 0

Figure. 2: Correlation between Total chemotherapy received (fraction of planned total dose) and the number of NaC cycles. Size reprezents number of patients. Boxplot by group. Control: no NAC cycles.

3.3 NAC ~ Leucocytes

## 
##  Spearman's rank correlation rho
## 
## data:  baza_de_date$`Neoadjuvant chemotherapy` and baza_de_date$`Lowest Leucocyte count`
## S = 20000, p-value = 0.1
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##    rho 
## -0.211

## 
##  Wilcoxon rank sum test with continuity correction
## 
## data:  Lowest Leucocyte count by Group
## W = 300, p-value = 0.4
## alternative hypothesis: true location shift is not equal to 0

Figure. 3: Correlation between Lowest Leucocyte count and the number of NaC cycles. Grade of Leucopenia highlighted. Boxplot by group. Control: no NAC cycles. Clinical Leucocyte tresholds marked by dotted blue line.

Figure. 4: Correlation between Lowest Leucocytes count and the number of NaC cycles by Stage (left) and Total chemotherapy received as a fraction of planned total dose (right). Clinical Leucocytes tresholds marked by dotted blue line. Statistics depict the Pearson correlation coefficient (with p-value) for the raw raltion, the partial corelaton by each covariate (with p-value) and the adjusted R-squared of a regression model with each covariate (with model F-statistic’s p-value).

3.4 NAC ~ Neutrophiles

## 
##  Spearman's rank correlation rho
## 
## data:  baza_de_date$`Neoadjuvant chemotherapy` and baza_de_date$`Lowest Neutrophile count`
## S = 20000, p-value = 0.1
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##    rho 
## -0.235

## 
##  Wilcoxon rank sum test with continuity correction
## 
## data:  Lowest Neutrophile count by Group
## W = 300, p-value = 0.3
## alternative hypothesis: true location shift is not equal to 0

Figure. 5: Correlation between Lowest Neutrophile count and the number of NaC cycles. Grade of neutropenia highlighted. Boxplot by group. Control: no NAC cycles. Clinical neutrophile tresholds marked by dotted blue line.

Figure. 4: Correlation between Lowest Leucocytes count and the number of NaC cycles by Stage (left) and Total chemotherapy received as a fraction of planned total dose (right). Clinical Leucocytes tresholds marked by dotted blue line. Statistics depict the Pearson correlation coefficient (with p-value) for the raw raltion, the partial corelaton by each covariate (with p-value) and the adjusted R-squared of a regression model with each covariate (with model F-statistic’s p-value).

3.5 NAC ~ Lymphocytes

## 
##  Spearman's rank correlation rho
## 
## data:  baza_de_date$`Neoadjuvant chemotherapy` and baza_de_date$`Lowest Lymphocyte count`
## S = 20000, p-value = 0.6
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##    rho 
## 0.0774

## 
##  Wilcoxon rank sum test with continuity correction
## 
## data:  Lowest Lymphocyte count by Group
## W = 200, p-value = 0.6
## alternative hypothesis: true location shift is not equal to 0

Figure. 6: Correlation between Lowest Lymphocytes count and the number of NaC cycles. Grade of lymphopenia highlighted. Boxplot by group. Control: no NAC cycles. Clinical lymphocyte tresholds marked by dotted blue line.

Figure. 4: Correlation between Lowest Leucocytes count and the number of NaC cycles by Stage (left) and Total chemotherapy received as a fraction of planned total dose (right). Clinical Leucocytes tresholds marked by dotted blue line. Statistics depict the Pearson correlation coefficient (with p-value) for the raw raltion, the partial corelaton by each covariate (with p-value) and the adjusted R-squared of a regression model with each covariate (with model F-statistic’s p-value).

3.6 NAC ~ Platelets

## 
##  Spearman's rank correlation rho
## 
## data:  baza_de_date$`Neoadjuvant chemotherapy` and baza_de_date$`Lowest Platelet count`
## S = 30000, p-value = 0.003
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##    rho 
## -0.423

## 
##  Wilcoxon rank sum test with continuity correction
## 
## data:  Lowest Platelet count by Group
## W = 400, p-value = 0.02
## alternative hypothesis: true location shift is not equal to 0

Figure. 7: Correlation between Lowest Platelets count and the number of NaC cycles. Grade of Thrombocytopenia highlighted. Boxplot by group. Control: no NAC cycles. Clinical Platelet tresholds marked by dotted blue line.

Figure. 4: Correlation between Lowest Leucocytes count and the number of NaC cycles by Stage (left) and Total chemotherapy received as a fraction of planned total dose (right). Clinical Leucocytes tresholds marked by dotted blue line. Statistics depict the Pearson correlation coefficient (with p-value) for the raw raltion, the partial corelaton by each covariate (with p-value) and the adjusted R-squared of a regression model with each covariate (with model F-statistic’s p-value).

3.7 NAC ~ Hemoglobin

## 
##  Spearman's rank correlation rho
## 
## data:  baza_de_date$`Neoadjuvant chemotherapy` and baza_de_date$`Lowest Hemoglobin`
## S = 30000, p-value = 0.0005
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##    rho 
## -0.484

## 
##  Wilcoxon rank sum test with continuity correction
## 
## data:  Lowest Hemoglobin by Group
## W = 400, p-value = 0.003
## alternative hypothesis: true location shift is not equal to 0

Figure. 8: Correlation between Lowest Hemoglobin value and the number of NaC cycles. Grade of anemia highlighted. Boxplot by group. Control: no NAC cycles. Clinical hemoglobin tresholds marked by dotted blue line.

Figure. 9: Correlation between Lowest Hemoglobin count and the number of NaC cycles by Stage (left) and Total chemotherapy received as a fraction of planned total dose (right). Clinical Hemoglobin tresholds marked by dotted blue line. Statistics depict the Pearson correlation coefficient (with p-value) for the raw raltion, the partial corelaton by each covariate (with p-value) and the adjusted R-squared of a regression model with each covariate (with model F-statistic’s p-value).

3.8 NAC ~ Creatinin, Urinary, Digestive, RT to lombo-aortics

Figure. 10: Creatinin toxicity, Urinary and Digestive Toxicity, RT to lombo-aortics by Neoadjuvant chemotherapy groups

4.1 RT given to LomboAortics

## Warning in if (is.na(.or.ci)) .or.ci = "" else .or.ci = paste0(" [",
## scales::number(limit(.or.ci), : the condition has length > 1 and only the
## first element will be used

## Warning in if (is.na(.rr.ci)) .rr.ci = "" else .rr.ci = paste0(" [",
## scales::number(limit(.rr.ci), : the condition has length > 1 and only the
## first element will be used

## Warning in chisq.test(t): Chi-squared approximation may be incorrect