Association of Sialyl T antigen with Cervical Cancer Lymph Node Status: An NRG Oncology/GOG study

Abstract

Objective:

Detection of lymph node metastases in cervical cancer patients is important for guiding treatment decisions, however accuracies of current detection methods are limited. We evaluated associations of abnormal glycosylation, represented by Tn and STn antigens on mucin (MUC) proteins, in primary tumor specimens with lymph node metastasis or recurrence of cervical cancer patients.

Methods:

Surgical specimens were prospectively collected from 139 patients with locally-advanced cervical cancer undergoing lymphadenectomy enrolled in a nation-wide clinical trial (NCT00460356). Of these patients, 133 had primary cervix tumor, 67 had pelvic lymph node (PLN) and 28 had para-aortic lymph node (PALN) specimens. Fixed tissue serial sections were immunohistochemically stained for Tn, STn, MUC1 or MUC4. Neuraminidase was used to validate Tn versus STn antibody specificity. Stain scores were compared with clinical characteristics.

Results:

Primary tumor STn expression above the median was associated with negative PLN status (p-value: 0.0387; odds ratio 0.439, 95% CI: 0.206 to 0.935). PLN had higher STn compared to primary tumor, while primary tumor had higher MUC1 compared to PALN, and MUC4 compared to PALN or PLN (p = 0.017, p = 0.011, p = 0.016 and p < 0.001, respectively). Tn and STn expression correlated in primary tumor, PALN, PLN, while Tn and MUC1 expression correlated in primary tumors (Spearman correlation coefficient [r] = 0.301, r = 0.686, r = 0.603 and r = 0.249, respectively).

Conclusions:

STn antigen expression in primary cervical tumors is a candidate biomarker for guiding treatment decisions and for mechanistic involvement in PLN metastases.

1. Introduction

Cervical cancer is considered a global health crisis [1]. Despite major improvements in detection of infection with the oncogenic human papillomavirus (HPV) causal agent and development of HPV vaccines, cervical cancer remains the most common gynecologic cancer and the leading cause of cancer-related deaths in women worldwide [2]. This is partly due to disparities in incidence, locally-advanced disease upon presentation and death rates in transitioning countries, and incomplete implementation of HPV screening and vaccine programs in transitioned countries [3]. A major factor affecting cervical cancer patient outcome is limitations in current standards for detecting lymph node metastases to guide treatment decisions. Cervical cancer patient prognosis and eligibility for clinical trials is affected by their cancer stage, which is determined according to updated International Federation of Gynecology and Obstetrics guidelines [4]. A current opinion is that more detailed classification of metastases is needed to optimize these guidelines [5].

Cervical cancer metastasizes primarily through the internal iliac, external iliac, obturator and common iliac pelvic lymph nodes (PLNs) [6]. Although para-aortic lymph node (PALN) metastases are much less common, their presence represents more advanced-stage disease and also defines the field of radiation exposure [7]. Knowing the pattern of lymph node spread before treatment could guide treatment strategies. Based on staging, cervical cancers are treated with various combinations of radiation and chemotherapy, including intensity-modulated radiotherapy (IMRT) modulation to maximize radiation dose to the target tissues and minimize exposure to adjacent normal tissues [8]. However, current technology has proven inadequate at detecting positive lymph nodes [9]. Development of novel imaging biomarkers could further optimize current imaging strategies [10].

Altered glycosylation of cell surface proteins and lipids is a common occurrence in cancer cells [11], and has been a frequent target for biomarker development and therapeutics [12]. A specific class of differences are a failure to assemble mature mucin-type O-glycans on mucins and other types of proteins [13, 14]. Mucin-type O-glycan assembly is initiated by the action of any of 20, in humans, different N-acetyl-α-D-galactosaminyltransferases (α-GalNAcTs) to transfer GalNAc from UDP-GalNAc to available Ser or Thr side chains usually in the Golgi apparatus [14]. The GalNAc is then subject to further modification by a variety of glycosyltransferases in a cell type specific manner to generate mature linear or branched O-glycans of up to seven or more monosaccharides [15]. A major pathway generates the core 1 Galβ1,3GalNAcα- disaccharide (referred to as the T antigen or T glycan), catalyzed by T-synthase encoded by the core 1 β3 galactosyltransferase (C1GALT1) gene. Failure to extend this disaccharide is a marker of some cancers [16]. In many cancers, the initial α-GalNAc also remains unextended, generating the so-called Tn antigen, owing to deficiencies in C1GALT1 or the X-linked Cosmc chaperone required for proper folding of C1GALT1 [15, 17]. In these cells, the Tn antigen is available for alternative modification by the sialyltransferase ST6GALNAC1 to form the NeuNAca2–6GalNAca1- disaccharide, or STn antigen [18]. The STn antigen can also result from overexpression of ST6GALNAC1 that competes with C1GALT1 [19]. Accumulation of Tn or STn has been associated with increased tumorigenic and metastatic behavior and immunosuppression [19, 20]. More specifically, the occurrence of Tn or STn glycans on the membrane-bound mucins MUC1 and MUC4 has been detected in multiple cancer types and implicated in metastatic potential [21, 22].

We previously demonstrated that Cosmc gene mutations are associated with appearance of Tn and STn antigens on cervical cancers [17]. This study hypothesized that positive staining for Tn, STn, MUC1 or MUC4 in primary tumor specimens predicts lymph node metastasis or disease recurrence. Specimens were prospectively collected from patients enrolled in the Gynecologic Oncology Group (GOG) clinical trial GOG0221: Glycoprotein and glycan profiling in patients with locally advanced cervical cancer (Stage IB2, IIA > 4 cm, IIB to IVA) undergoing pelvic and para-aortic (abdominal) lymphadenectomy.

The occurrence of Tn or STn glycans on the membrane-bound mucins, MUC1 and MUC4, has been detected in multiple cancer types and implicated in metastatic potential [21, 22]. This study hypothesized that positive staining for Tn, STn, MUC1 or MUC4 in primary tumor specimens predicts lymph node metastasis or disease recurrence. Specimens were prospectively collected from patients enrolled in the Gynecologic Oncology Group (GOG) clinical trial GOG0221: Glycoprotein and glycan profiling in patients with locally advanced cervical cancer (Stage IB2, IIA > 4 cm, IIB to IVA) undergoing pelvic and para-aortic (abdominal) lymphadenectomy.

2. Materials and methods

2.1. GOG0221 Clinical Trial

This project (NRG GY-TS012 project) received IRB, NRG Oncology and National Cancer Institute approval to evaluate specimens from the clinical trial GOG0221. The primary objective was to determine whether the presence of a mutation in T-synthase or Cosmc and/or the presence of positive immunohistochemical expression of Tn antigen or STn antigen in tumor specimens is associated with progression-free survival (PFS) or overall survival (OS). Patients with stage IB2, IIA>4cm, IIB, III, or IVA cervical cancer undergoing pelvic and para-aortic (abdominal) lymphadenectomy were eligible for this trial, which did not meet its specimen accrual goal before closure. A secondary objective was to determine whether Tn or STn antigen expression in tumors is associated with lymph node metastasis, local control or disease recurrence/progression. GOG0221 enrolled 159 patients including 5 patients who were ineligible due to either wrong stage, second primary, wrong cell type, wrong primary or inadequate pathology. Specimens from 139 study eligible patients were distributed for this study objective. All participants signed written informed consent for GOG-0221.

2.2. GOG0221 Clinical Trial Specimens

Five types of surgical specimens were collected, fixed in formaldehyde and embedded in paraffin (FFPE): primary cervix tumor and positive para-aortic, obturator, external iliac and common iliac lymph nodes. The obturator, external iliac and common iliac lymph nodes were all categorized as PLN. Overall, 133 patients had primary cervix tumor specimens, 28 patients had PALN specimens (26 of which were confirmed by pathology review as positive for cancer/PALN positive) and 67 patients had PLN specimens (66 of which were confirmed by pathology review as positive for cancer/PLN positive). Seven FFPE sections were cut and shipped to D.M.B at the Stephenson Cancer Center, University of Oklahoma Health Sciences Center. Each cancer positive specimen was stained for MUC1, MUC4, Tn and STn antigen using optimized and validated assays.

2.3. Antibodies

Antibodies and sources are listed in Supplemental Table 1. Monoclonal antibodies to Tn and STn were validated and provided as tissue culture supernatants by Dr. U. Mandel (University of Copenhagen, Denmark).[23] The commercial antibodies were generated using peptides, and not glycopeptides, of MUC1 and MUC4.

2.4. Biomarker Analysis of GOG0221 Clinical Specimens

All immunohistochemical assays were performed using Leica Bond-III^™ Polymer Refine Detection system (DS 9800). Deparaffinization and rehydrated FFPE sections were subjected to target retrieval (100°C for 20 minutes in citrate buffer, pH 6.0), incubated with 5% goat serum for 30 minutes and then treated with a peroxidase-blocking reagent. Sections were then incubated primary antibody for 60 minutes followed by incubation with secondary antibody, post-primary IgG-linker and/or Poly-HRP IgG reagents, and detection with 3, 3′-diaminobenzidine tetrahydrochloride and hematoxylin counter stain. Positive and negative (omission of primary antibody) controls were stained in parallel. For Tn and STn, two set of slides were evaluated in parallel for each specimen. One set was treated with 50 mU/ml neuraminidase from Arthrobacter ureafaciens (Roche, 10269611001) in 10 mM Tris-HCl, pH 5.5 for 2.5 hours in a humid chamber at 37°C before the antigen retrieval.

2.5. Tn and STn Biomarker Assay Validation

Sections of colon from the T-synthase knockout (KO) mouse (villin::Cre/floxed C1GalT1, gift of Lijun Xia, Oklahoma Medical Research Foundation, who cared for the mice in accordance with institutional guidelines) which express increased levels of Tn and STn [24] served as positive controls. Colon sections from C57BL/6J mice wild type (WT) mice, which express minimal Tn and STn served as negative controls. Neuraminidase, an enzyme that cleaves Sialyic acid from STn, was used to verify STn and Tn antibody specificity via neuraminidase reduction of staining of STn positive controls and lack of neuraminidase effect on staining of Tn positive controls, respectively.

2.6. Biomarker Expression

Experienced gynecologic pathologists reviewed the immunohistochemical staining of blinded clinical specimen and control slides. The STn and Tn stains were reviewed by R.C. for stain intensity on a scale of 0 = no stain, 1 = weak stain, 2 = moderate stain, and 3 = strong stain and the percent positively stained cells on a scale of 0 through 100%. The H-score was calculated by multiplying the intensity score by the percent positive cells resulting in a scale of 0 through 300. MUC1 and MUC4 stains were reviewed by S.H. using the same intensity scale, and the percent positively stained cells were rated on a scale of 1 through 4, with 1 = 0 – 25% positive, 2 = 26–50% positive, 3 = 51–74% positive and 4 = 75–100% positive. The MUC1 and MUC4 H scores were calculated by multiplying the intensity score by the percent positive score resulting in a scale of 0 through 12. The patterns of biomarker expression are shown in the accompanying Data in Brief [25].

2.7. Data Analysis

The data set used for baseline characteristics and survival information was retrieved from NRG INGRES data base as data present on 4/4/2019. Baseline characteristics collected were age, ethnicity, race, performance status, tumor histology, tumor grade, cancer stage, PALN and PLN status, clinical tumor size and local control. Local control was defined as success if a patient had no progression or recurrence in the pelvic field within three months of completing first line chemoradiation, otherwise defined as failure. The biomarker expression in PLNs was calculated as a simple average of the expression in obturator, external iliac and common iliac lymph nodes available for each patient.

2.8. Statistical Analysis

Associations for each biomarker (MUC1, MUC4, Tn, STn, and combination of STn and Tn) by specimen type (primary tumor, PALN, PLN) with PALN, PLN or local control status were tested by Fisher exact tests using biomarker expression dichotomized by the H score sample median. Also, associations of PALN status, PLN status, or local control with each baseline characteristic were explored. In general, Monte-Carlo based exact chi-square tests or Fisher exact tests were used for a discrete-type variable, Monte-Carlo based exact Spearman’s rank correlation coefficient tests were performed when a discrete-type variable had ordinal features, and a simple logistic regression was used for an interval-type variable. Log-rank tests were used to explore associations of H scores dichotomized by sample median or baseline characteristics with PFS and OS. Cox proportional hazards (PH) model was used to estimate hazard ratio (HR) and corresponding confidence interval (CI). Odds ratio (OR) and corresponding CI were estimated by logistic model.

A logistic or Cox PH multiple regression were utilized to further explore associations between explanatory variables and a response variable whenever this response variable had more than one explanatory variable with a marginal p-value less than 0.1. Backward selection method was used for model building.

Pairwise comparisons for each biomarker expression among primary tumor, PALN, and PLN were examined by Wilcoxon signed rank tests. In addition, pairwise correlations among the biomarkers of MUC1, MUC4, Tn and STn by tissue type were explored by Spearman’s rank correlation coefficient. All observations with missing values including unknown or not reported or not available were excluded in statistical analyses. A significance level was set at 0.05. There was no adjustment for multiple testing. SAS 9.4 was used in the analyses.

3. Results

3.1. Clinical Trial

Between April 2, 2007 and July 16, 2016, 159 Stage IB2 to IVA cervical cancer patients from GOG Foundation Institutions across the US were enrolled in the GOG0221 trial. Cut FFPE sections of primary, PALN and PLN specimens from 139 patients who met eligibility criteria were evaluated in this study. Baseline characteristics, PALN status, PLN status and local control status for these patients are listed in Tables 1 and 2.

TABLE 1.

Patient Characteristics

Characteristic	Number (%)
Age (years)
Median	47·7
Min - Max	21·1 – 85·6
Q1 - Q3	40·1 – 56·8
20 – 29	8 (5·8)
30 – 39	25 (18·0)
40 – 49	49 (35·3)
50 – 59	32 (23·0)
60 – 69	22 (15·8)
70 – 79	2 (1·4)
80 – 89	1 (0·7)
Ethnicity
Hispanic	16 (11·5)
Non-Hispanic	76 (54·7)
Unknown/Not Reported	47 (33·8)
Race
Unknown/Not Reported	10 (7·2)
Asian	7 (5·0)
Black/African American	13 (9·4)
American Indian/Alaskan	2 (1·4)
Native Hawaiian/Pacific Islander	1 (0·7)
White	106 (76·3)
Performance Status
0	131 (94·2)
1	7 (5·0)
2	1 (0·7)
Not Graded	5 (3·6)

TABLE 2.

Tumor Characteristics

Tumor Characteristic	Number (%)
Histology
Adenocarcinoma, Unsp.	10 (7·2)
Endometrioid Adenocarcinoma	2 (1·4)
Mixed Epithelial Carcinoma	1 (0·7)
Adenosquamous	6 (43)
Squamous Cell Carcinoma	118(84·9)
Villoglandular Adenocarcinoma	1 (0·7)
Serous Adenocarcinoma	1 (0·7)
Tumor grade
1	7 (5·0)
2	66 (47·5)
3	61 (43·9)
Not Graded	5(3·6)
Stage
I	52 (37·4)
II	63 (45·3)
III	21 (15·1)
IV	3 (2·2)
Local control
Failure	21 (15·1)
Success	117 (84·2)
NA	1 (0·7)
Clinical tumor size
Median	5·5
Min - Max	0 – 10
Q1 - Q3	4·5 – 6·5
N Missing	0
Histology
Adenocarcinoma, Unsp.	10 (7·2)
Endometrioid Adenocarcinoma	2 (1·4)
Mixed Epithelial Carcinoma	1 (0·7)
Adenosquamous	6 (4·3)
Squamous Cell Carcinoma	118 (84·9)
Villoglandular Adenocarcinoma	1 (0·7)
Serous Adenocarcinoma	1 (0·7)
Reviewed pelvic lymph node status
Negative	57 (41·0)
Positive	76 (54·7)
NA	6 (4·3)
Reviewed para-aortic lymph node status
Negative	98 (70·5)
Positive	29 (20·9)
NA	12 (8·6)

3.2. Associations of Survival

Associations of PFS or OS with each biomarker by specimen type, baseline characteristics, PALN status, or PLN status were evaluated whenever it was appropriate and feasible. No statistically significant evidence was found to support associations between the biomarkers and PFS or OS. As anticipated, PFS was separately significantly associated with stage, clinical tumor size (dichotomized by median), and PALN status (Fig. 1). Patients with early stage cervical cancer had longer PFS compared to patients with advanced stage cervical cancer (p-value [p]: 0.0079; HR [95% CI]: 0.403 [0.209 – 0.783] for stage I to stage III/IV, and 0.444 [0.244 – 0.83] for stage II to stage III/IV). Patients with smaller clinical tumor sizes had longer PFS compared to patients with larger clinical tumor sizes (dichotomized by median, p: 0.0102; HR: 0.5192; 95% HR CI: 0.308 – 0.857). Patients with negative PALN had better PFS than patients with positive PALN (p-value by permutation-based log-rank test: 0.0044; HR: 0.45; 95% HR CI: 0.264 to 0.791.

Fig 1.

Kaplan-Meier curves. A associations of PFS with PALN status (A), clinical tumor size (B) and clinical stage (C).

OS was significantly associated with stage, clinical tumor size (dichotomized by median), and PALN status (Fig. 2). Patients with early stage cervical cancer had better OS compared to patients with advanced stage cervical cancer (p: 0.0171; HR [95% CI]: 0.452 [0.226 – 0.917] for stage I to stage III/IV, 0.417 [0.218 – 0.821] for stage II to stage III/IV). Patients with smaller clinical tumor sizes had better OS compared to a patient with larger clinical tumor sizes (p: 0.0113; HR: 0.497; 95% HR confidence interval [CI]: 0.281 – 0.855). Patients with negative PALN had better OS than patients with positive PALN (p-value by permutation-based log-rank test: 0.0002; HR: 0.349; 95% HR CI: 0.196 to 0.633).

Fig 2.

Kaplan-Meier curves. A associations of OS with PALN status (A), clinical tumor size (B) and clinical stage (C).

In Cox PH multiple regression, candidate explanatory variables included primary tumor MUC4, stage, clinical tumor size and PALN status for PFS, and primary tumor combined STn and Tn H score, stage, clinical tumor size and PALN status for OS. Based on parsimony principle and model fitness criterion, PFS was significantly associated with stage and PALN status. The HR for negative PALN status was 0.406 compared to positive PALN status with a 95% CI of 0.237 to 0.717 with adjustment for stage. The HR for stage I to stage III/IV was 0.308 with a 95% CI as 0.150 to 0.631, and the HR for stage II to stage III/IV was 0.408 with a 95% CI as 0.221 to 0.779 adjusted for PALN status. Similarly, OS was significantly associated with stage and PALN status. The HR for negative PALN status was 0.302 compared to positive PALN status with a 95% CI as 0.181 to 0.59 after adjustment for stage. The HR for stage I to stage III/IV was 0.342 with a 95% CI as 0.158 to 0.736, and the HR for stage II to stage III/IV was 0.396 with a 95% CI as 0.203 to 0.799 after adjustment for PALN status.

3.3. Biomarker Associations with PALN, PLN or Local Control Statuses

For each run, staining specificity was confirmed by lack of effect of neuraminidase treatment on Tn staining, and neuraminidase reduction of STn staining in positive and negative control specimens (Fig. 3) and all clinical specimens evaluated (examples in Fig. 4). The H scores of the staining are summarized in Table 3. Fisher’s exact test supported primary tumor STn association with PLN status (p=0.0387). The medians for primary tumor STn H-scores in patients with PLN positive versus negative specimens were 50 versus 95, and the means for primary tumor STn H-scores in patients with PLN positive versus negative specimens were 80 vs 103. The odds ratio of having a positive PLN was 0.439 (95% CI: 0.206 to 0.935) for a patient with higher primary tumor STn expression compared to a patient with lower primary tumor STn expression. This indicates that patients with lower STn staining in the primary tumor have a higher risk of having positive PLN. There were no significant associations between PALN, PLN, or local control status with any of the other biomarkers examined in this study. Patients with positive PALN status were more likely to have worse performance status (p=0.0069), and patients with early stage were more likely to have success in local control (p=0.0449). The Fisher exact test suggested trends (p-values between 0.05 and 0.1) for associations between PLN status with dichotomized age or histology, and histology with local control status.

Fig 3.

Validation of Antibody Specificity using T-Synthase KO (Tn ad STn positive) and WT (STn and Tn negative) colon section controls. Immunohistochemical staining of the colon tissues was conducted with anti-Tn mAbs (BAG6, IE3, and 5F4) and the anti-STn mAb (TKH2) in the without (−NA) ot with (+NA) of neuraminidase treatment. The red circle outlines STn positivity in the T-synthase KO colon section.

FIG 4.

Example of Tn and STn antibody specificity verification in clinical specimen with similar levels of Tn and STn expression. Sections were stained with Tn or STn antibodies in the absence (−NA) or presence of (+NA) neuraminidase treatment.

TABLE 3.

Biomarker expression Immunohistochemistry stain in the 139 patients by biomarker and specimen

	N= 139
Expression IHC H-score in primary cervix tumor	MUC1	MUC4	Tn	STn	Combination of Tn and STn
Median	9	0	105	81	100·5
Min - Max	0 – 12	0 – 12	1 – 290	0 – 279	1 – 290
Q1 - Q3	3 – 12	0 – 3	65 – 137	18 – 159	58 – 120
Num. of observations with missing value	6	6	17	20	17
Expression IHC H-score in para-aortic lymph node
Median	3	0	102·5	120	96
Min - Max	0 – 12	0 – 12	0 – 252	0 – 260	0 – 255
Q1 - Q3	3 – 105	0 – 0	51 – 115	13 – 190	70 – 108
Num. of observations with missing value	111	111	113	113	113
Expression IHC H-score in pelvic lymph node by average
Median	6	0	105·5	115	100·5
Min - Max	0 – 12	0 – 12	1 – 240	0 – 275	1 – 215
Q1 - Q3	3·5 – 9	0 – 0	62·3 – 143·6	22 – 166·2	47·2 – 123
Num. of observations with missing value	72	72	75	75	75
Difference of expression IHC H-score between primary and PALN
Median	2	0	11.5	0	165
Min - Max	−9 – 12	−12 – 12	−95 – 102	−259 – 162	−95 – 117
Q1 - Q3	0 – 9	0 – 6	−9 – 49	−52 – 50	−2 – 34
Num. of observations with missing value	112	112	117	118	117
Difference of expression IHC H-score between primary and PLN
Median	0	0	−4	−9·5	2
Min - Max	−12 – 12	−6 – 12	−130 – 139·7	−274 – 111	−102 – 146·3
Q1 - Q3	−1·5 – 3	0 – 5	−29 – 35	−54 – 125	−18 – 30·5
Num. of observations with missing value	78	78	84	85	84
Difference of expression IHC H-score between PALN and PLN
Median	−0·25	0	0	3·8	−3
Min - Max	−9 – 3	−6 – 12	−122 – 72	−160·5 – 90	-97 – 95
Q1 - Q3	−3·5 – 0·5	0 – 0	−30·3 – 2·5	−15 – 22·5	−30·3 – 3·3
Num. of observations with missing value	119	119	120	121	120
Expression IHC H-score in external iliac LN
Median	8	0	103	119	100
Min - Max	0 – 12	0 – 12	1 – 230	0 – 295	1 – 215
Q1 - Q3	3 – 12	0 – 0	70 – 135	17 – 160	35 – 126
Num. of observations with missing value	98	98	102	101	102
Expression IHC H-score in obturator LN
Median	6	0	107·5	115	102·5
Min - Max	0 – 12	0 – 6	1 – 240	0 – 270	1 – 195
Q1 - Q3	3 – 12	0 – 0	55 – 145	31 – 166	41 – 125
Num. of observations with missing value	95	97	97	97	97
Expression IHC H-score in common iliac LN
Median	6	0	121	159	105
Min - Max	0 – 12	0 – 3	1 – 245	0 – 290	1 – 210
Q1 - Q3	3 – 9	0 – 0	90 – 154	27 – 197	98 – 114
Num. of observations with missing value	116	116	118	116	118

LN = lymph node; IHC = immunohistochemistry; Q1=first quartile; Q3=third quartile.

Logistic multiple regressions were used to study association of PLN status with local control. The candidate explanatory variables for PLN status included primary tumor STn, age and histology, and the candidate explanatory variables for local control were histology and stage. Based on parsimony principle, model fitness criterion and a significance level of 0.05, only primary tumor STn was significantly associated with PLN status. Neither histology or stage was significantly associated with local control.

3.4. Associations Among Biomarkers

Pairwise comparisons of each biomarker expression among primary tumor, and metastatic tumor in the PALN and PLN showed significantly higher STn expression in the PLN compared to the primary tumor (Wilcoxon signed rank sum test p-value: 0.017). MUC1 expression was significantly higher in the primary tumors compared to PALN (p = 0.011), but not compared to the PLN specimens. MUC4 expression was significant higher in the primary tumor compared to either PALN or PLN specimens (p = 0.016 and p <0.001, respectively).

Spearman’s rank correlation coefficient (r) indicated that Tn expression correlated with STn expression in the primary tumor (r = 0.301, p = 0.001), PALN (r = 0.686, p < 0.001) or PLN (r = 0.603, p < 0.001), Tn expression in the primary tumor correlated with MUC1 expression in the primary tumor (r = 0.249, p = 0.0057), and MUC1 expression correlated with MUC4 in PLN (r = −0.271, p = 0.0267). No other comparisons were found to be statistically significant.

4.0. Discussion

This evaluation of Tn, STn, MUC1 and MUC4 in cervical cancer specimens identified significant association of low STn antigen staining in primary tumors with the absence of positive PLNs. The lack of association of STn staining with PALN suggests that STn may play more predominant role in the early metastases process associated with PLN positivity, as opposed to later stage disease associated with PALN positivity [7]. The significantly higher STn expression observed in PLN compared to primary tumors may reflect migration of STn positive cells from the primary tumor to PLNs. In contrast, significantly higher MUC1 and MUC4 proteins were observed in the primary tumors compared to PLN or PALN. These observations, combined with the fact that MUC1 and MUC4 were assayed independently of their glycosylation status, suggest it is the glycosylation status of these proteins that determines their ability to drive metastases. Consistent with this, circulating anti-glycan antibodies predicted improved PFS and disease specific survival of cervical cancer patients receiving external beam radiation therapy and brachytherapy, but not for patients receiving external beam radiation therapy alone [26]. The improved survival may be due to the ability of anti-glycan antibodies to neutralize tumor glycans. PFS and OS associations with clinical factors known to influence patient survival support the validity of this study’s data. The lack of association between biomarker staining with PFS or OS in this study may be due to the closure of this study before its accrual goals were achieved.

This study differs from other published reports in that it prospectively collected specimens and data to determine if Tn and STn status of the primary tumor is associated with lymph node status, included PALN specimens and utilized neuraminidase to validate Tn and STn specificity of validated antibodies, whereas the other studies were retrospective, did not include PALN specimens and used either lectins or antibodies without neuraminidase validation to detect Tn and STn [27–29]. In comparison to the other published reports, our study provides more robust data in that our immunohistochemical staining was validated to be specific for STn by reduction of positive signal, and Tn by lack of reduction of positive signal, upon neuraminidase treatment. Furthermore, inclusion of PALN specimens in our study provides novel information not included in the other studies.

These previous studies consistently found STn and Tn expression in cervical cancer, but absent, or much lower, levels in normal cervical tissue. Findings in these studies were inconsistent findings with regard to the prognostic significance of STn and Tn, and are likely due to a combination of a variety of factors including the different methodologies and numbers and types of specimens evaluated. A study using lectins to detect Tn antigen in 111 cervical cancers found that Tn expression correlated with parametrial spread, positive lymph node status and worse 5-year survival [29]. Another study used HB-Tn1 and B1.1 Tn antibodies and HB-STn1 and TKH2 STn antibodies to stain 83 invasive squamous cell carcinomas with 27 matching positive PLNs, 22 severe dysplasia, 24 carcinomas in situ, and 36 normal cervical squamous epithelia [28]. This study found STn, but not Tn, expression in severe dysplasia and carcinomas in situ, no significant differences in Tn or STn expression patterns between primary tumors and positive PLNs, and no associations of Tn or STn expression with clinical characteristics. A study that used TKH2 mAB to detect STn in 24 normal cervical tissues and 53 carcinomas including 36 matched positive lymph nodes found no significant association of STn staining with presence of metastasis [27].

The results of this study support further development of STn as a biomarker to add to the armamentarium of strategies for clinical staging of cervical cancer patients. Low STn staining was associated with positive PLN status, which occurs in earlier phases of cervical cancer spread, but not with PALN status, which occurs in later stage disease. None of the other biomarker endpoints evaluated showed significant associations with clinical endpoints, however Tn and STn staining levels correlated with each other. Observations of known associations of clinical endpoints with survival in this study support the validity of the data and analysis methodology.

Highlights.

• Sialyl T antigen and T antigen expression positively correlated in both primary tumors and lymph node metastases specimens.

• Pelvic lymph node metastases had higher sialyl T antigen expression in comparison to the primary cervical tumors.

• Elevated sialyl T antigen in primary cervical tumors correlated with absence of pelvic lymph node metastasis in patients

Data Sharing

All data elements of the individual participants that are required to reproduce results reported in this article (text, tables, figures, supplemental material) after de-identification will be made available. Data must comply with rules and regulations of the NCTN Data Archive. The protocol document and data dictionary will also be made available. The data can be accessed at the National Institutes of Health, National Cancer Institute, NCTN/NCORP Data Archive https://nctn-data-archive.nci.nih.gov/. The documents can be access at the National Institutes of Health, National Cancer Institute, NCTN/NCORP Data Archive https://nctn-data-archive.nci.nih.gov/. Data and documents will be submitted to the NCTN/NCORP Data Archive within 6 months of publication of this article. Data will be made available to researchers with an approved Data Use Agreement. Data will be made available for researchers who wish to analyze the data in secondary studies to enhance the public health benefit of the original work. Data requesters must sign a Data Use Agreement before being able to download data for a given data request. Researchers must adhere to all terms of access in the Data Use Agreement. The Data Use Agreement is in effect for up to 3 years. An extension can be pursued or the data in all forms must be destroyed.