Combined score of Y chromosome loss and T-cell infiltration improves UICC based stratification of esophageal adenocarcinoma

Background Staging, especially clinical lymph node staging in esophageal adenocarcinoma has only moderate sensitivity and specificity. Therefore, we evaluated combined molecular markers to predict prognosis. Patients and methods 890 tumor tissue samples were obtained from patients who underwent surgery for esophageal adenocarcinoma with curative intent. These were stained by tissue micro array for 48 markers which are associated with tumorigenesis and correlated with clinical data (TNM-staging, overall survival) by multivariate Cox regression. Results Two markers (preserved Y chromosome and high grade of (CD3+) T-cell infiltration) were found to be significantly and independently associated with better overall survival. We formed a score (called CY score) from the two markers. The more markers are positive and thus the higher the score (ranging from 0 to 2), the better the overall survival, independently of UICC. Moreover, we developed a combination score of the UICC and CY score based on cluster analysis. Patients with a UICC stage of III with the presence of both traits (CY=2) can be assigned to a better prognosis group (group II), whereas patients with a UICC stage of I without both traits (CY=0) must be assigned to a worse prognosis group (group II). Therefore, patients in stage I with adverse molecular signature might benefit of multimodal therapy. Conclusion In summary, the CY score adds prognostic information to the UICC stage based on tumor biology in esophageal adenocarcinoma and warrants further evaluations in independent clinical cohorts.


Introduction
Esophageal cancer is a common cancer with a high mortality ranking seventh in terms of incidence and sixth in mortality worldwide (1).In developed countries the incidence of esophageal cancer is rising (2).Whereas very early in the disease endoscopic treatment can lead to cure, surgical resection often combined with chemotherapy or radiochemotherapy was established for locally advanced tumors.Even with these intensive therapeutic regimens the overall survival of patients with esophageal cancer remains poor (3,4).More recently, also immunotherapy has been introduced in the curative first line setting (5).
The staining was evaluated and quantified by a pathologist and the open-source software QuPath and classified as negative or positive staining for the marker.The two evaluation strategies were compared with very high concordance of results.In the case of discordance, the pathologist determined the division into negative/positive.
Concerning CD3 cell infiltration into the tumor, we used two different evaluation strategies: a semiquantitative method in which a pathologist (AQ) estimated the extent of T lymphocytes in the stroma and divided the extent into the two groups "low" and "high" and a digital method using the freely available software QuPath.QuPath standardized the amount of CD3 positive cells in the tissue to 1 mm 2 in absolute numbers, the median was then taken to divide into the two groups "low" and "high".This was then compared to the primary semiquantitative assessment with very high concordance of results.In the case of discordance, the pathologist determined the division into "low" and "high".
Clinical data (especially age, sex, survival time, survival status, last follow-up, date of surgery, whether neoadjuvant therapy had been given, state at the time of surgery (pathological tumor extent (y)pT, pathological nodal state (y)pN, clinical metastasis state (cM) and UICC-stage) were collected prospectively according to a standardized protocol.

Statistical analysis
For statistical analysis, IBM SPSS Statistics for Windows (Version 27) was used.
First, univariate Cox regression was used to test the correlation of each marker with survival time.Significant markers with a pvalue (according to Bonferroni correction for multiple comparison) ≤ 0.001 were tested in multivariate Cox regression.A score (called CY score) was formed from the two remaining markers (detection of preserved Y chromosome and high grade of CD3 cell infiltration).A score value of 0 indicates that no marker is positive (loss of Y chromosome and low CD3 cell infiltration), 1 indicates that one marker is positive (either preserved Y chromosome or high CD3 cell infiltration), and 2 indicates that both markers are positive (preserved Y chromosome and high grade of CD3 cell infiltration).The calculation of the score was possible in 620 cases since there staining for both markers was available.To test the correlation of each marker (detection of preserved Y chromosome and CD3 cell infiltration) as well as of the CY score with survival time, a Kaplan-Meier curve was calculated comparing groups using the log-rank test.In addition, a multivariate Cox regression including UICC was calculated.In the next step, the correlation of the markers (detection of preserved Y chromosome and CD3 cell infiltration) and CY score with (y)pT, (y)pN and (y) UICC were tested by calculating the Spearman coefficient and cross tabulation, respectively.
To check for CY score validity in subgroups, Cox regression of the CY score was calculated for patients with a lower versus higher UICC stage (UICC stage 1 and 2 versus 3 and 4), for patients having undergone neoadjuvant chemotherapy or not, for younger versus older patients (< versus ≥ 65 years) and for male versus female patients.
Furthermore, the UICC stage was combined with the CY score.Therefore, the strata of the UICC stage were substratified based on the three groups of the CY score.Using cluster analysis with Ward's method and squared Euclidean distance as the proximity measure, these were reassembled into new groups of at least two substrata based on their similarity with respect to 0.5-, 1-, 1.5-, 2-, 2.5-, 3-, 3.5-, 4-, 4.5-, 5-, 5.5-, 6-, 6.5-and 7-year survival.To test the discriminatory power of the new CY-UICC score, a Kaplan-Meier curve was calculated comparing groups using the log-rank test.
All tests were two-sided; p values < 0.05 were considered statistically significant.Significance was marked as follows: * for p ≤ 0.05, ** for p ≤ 0.01 and *** for p ≤ 0.001.

Preserved Y chromosome and CD3 cell infiltration in the tumor are the most significant markers correlating with longer overall survival
As a first step, we examined which of the 44 markers correlated with prolonged overall survival (Table 2).
Then, a multivariate Cox regression was performed on the markers that correlated significantly with prolonged overall survival.The markers high CD3 cell infiltration into the tumor (p = 0.003, HR 0.69) and preserved Y chromosome (p = 0.005, HR 0.72) were found to stay significant and remained significant after adding UICC to the multivariate survival analysis.

A score of preserved Y chromosome and high CD3 cell infiltration in the tumor correlates with better overall survival
A score was formed from these two markers.Here, 0 corresponds to no positive marker (loss of Y chromosome and low infiltration of CD3 cells), 1 to one positive marker, and 2 to both markers being positive (Y chromosome preserved and high CD3cell infiltration into the tumor).Higher CY scores were significantly associated with better survival (CY=0 vs. 1 p=0.0002,CY=1 vs. 2 p=0.00001) (Figure 1).This was also true when controlling for UICC stage in a multivariate COX regression (p=0.00007).Correlation of the individual markers and the CY score with other clinical parameters demonstrated an association of high CD3 cell tumor infiltration and preserved Y chromosome with longer survival, a low pN and a low UICC.A high CY score additionally correlates with a smaller pT.All these associations are only to be seen in treatment-naïve patients (see Table 3).Histologic grading, treatment effect (Cologne regression score), lymphovascular invasion and margin status did not show a significant association with the CY score.
The CY score is prognostic for all subgroups but for female patients Concerning the validity of the CY score for prognosis of survival in subgroups, it is valid for patients with low (p=0.039)versus high (p=0.00016)UICC stage, for patients having undergone neoadjuvant therapy (p=0.002) or not (p=0.0003)and for younger (<65 years) (p=0.001) and older patients (p=0.00006).Concerning female patients, a subgroup analysis was not possible since in the analyzed cohort the score had been able to be calculated in too few women (n=28).
The CY score in combination with the UICC stage identifies patients with a better or worse prognosis based on tumor biology Substratifying the UICC stage using the three groups of the CY score and clustering these substrata by survival over time results in a new combination score.This newly formed combination score (CY-UICC score) is composed according to Figure 2.
The CY-UICC score significantly separates four different prognostic groups (group 1 vs. 2 p= 0.007, 2 vs. 3 p=0.001, 3 vs. 4 p=0.000000002)(see Figure 3).Median overall survival in the groups is 70.2 months (CY-UICC score 1), 42.5 months (CY-  Clustering of the CY-UICC score groups shows that considering the CY score, patients with a UICC stage of 3 with the presence of both traits (CY=2) can be assigned to a better prognosis group (group 2), whereas patients with a UICC stage of 1 without both traits (CY=0) must be assigned to a worse prognosis group (group 2) (see Figure 2).Figure 4 compares overall survival of the newly developed CY-UICC score and UICC.

Discussion
We have shown that a preserved Y chromosome and high CD3 cell infiltration in the tumor are important markers for prolonged overall survival in esophageal adenocarcinoma.
Y chromosome loss (LoY) is associated with aging in men and occurs more frequently in men who smoke (21).LoY is associated with a higher risk for developing non-hematological cancer in general (22).Specifically, we have shown that LoY is a common phenomenon in esophageal adenocarcinoma and is associated with shorter overall survival (8).The functional effect leading to decreased survival in esophageal cancer has not been fully elucidated until yet.It might be a marker for general chromosomal instability and is detectable already in precursing Barrett´s esophagus (23).Furthermore, LoY was shown to be associated with an epigenetic phenotype with methylation of genes being involved in cell proliferation and cell cycle regulation (21).Additionally, the loss of function of genes of the nonrecombinant region of Y were shown to play a role in cell cycle regulation and behave as dosage-sensitive tumor suppressors (24).Clustering groups to form the CY-UICC score.Tumor infiltrating lymphocytes (TIL) are involved in immune editing of tumor cells.By elimination of (immunogenic) tumor cells, they reduce tumor growth and at the same time support the eventual outgrowth of less immunogenic tumor cells by selection pressure (25,26).
Overall, high lymphocyte infiltration of the tumor is associated with longer survival.CD3 cell tumor infiltration is known as good prognostic factor concerning survival in ductal breast cancer (27), nonsmall cell lung cancer (NSCLC) (28), intestinal-type gastric cancer (29), hepatocellular carcinoma (HCC) (30), gastrointestinal stromal tumor (GIST) (31), ovarian cancer (32), bladder cancer (33), oral squamous cell carcinoma (34) and nasopharyngeal carcinoma (35).For colon cancer, it was even shown that immune cell density in the tumor was of more prognostic importance than UICC classification (36).In this paper, we have shown that using the CY score adds prognostic information to the UICC stage, most probably by resembling tumor biology (especially immunogenicity and genetic stability).Limitations of the study are its monocentric and retrospective character, a strength its large sample size for this comparably rare type of cancer.Staging, especially clinical lymph node staging is difficult, as sensitivity and specificity of lymph node detection in CT scans and endoscopic ultrasound have low sensitivity (52 -81%) and specificity (73 -87%) (37-39).As guidelines recommend multimodal therapy for patients with esophageal cancer in UICC stage II/III, lymph node staging is performed even though diagnostic techniques have limitations.
Here, we show that patients in UICC stage I with adverse molecular signature (CY score 0) clinically behave like UICC stage II patients and might benefit from multimodal therapy.UICC stage III patients with a favorable tumor biology (CY score 2) have a better prognosis compared to UICC stage III patients with a less favorable tumor biology.UICC stage IV patients with a good tumor biology (CY score 2) are a very rare event (CY-UICC=2 n=8 vs. CYUICC=1 n=46 and CYUICC=0 n=48) and in these cases the protective effect of the good tumor biology probably cannot outweigh the adverse stage IV cancer features.
In summary, the CY score adds prognostic information to the UICC stage and warrants further evaluations in independent clinical cohorts.

FIGURE 3
FIGURE 3Association of CY-UICC score and overall survival.

FIGURE 4
FIGURE 4Association of CY-UICC score vs. UICC and overall survival.The groups of CY-UICC score 4 and UICC 4 are identical.

TABLE 1
Clinical data of tumor samples.

TABLE 2
Marker: analyzable TMAs (n), mean of marker expression between 0 (negative) and 1 (positive), p value and hazard ratio of univariate Cox regression of the marker and overall survival.

TABLE 2 Continued
Association of CY score with overall survival.

TABLE 3
Association of markers with clinical data.