Optimization Of Processing Preoperative Breast Cancer Axillary Staging

BackgroundAxillary lymph ndoe (ALN) is the initial and most common metastatic site of breast cancer. It is the most important prognostic factor of early stage breast cancer. Precise axillary staging is the crucial guidance for choice of local treatment, decision of systemic multidisciplinary treatment and prognosis. Sentinel lymph node biopsy (SLNB) is a minimal invasive biopsy technique of precise axillary staging. Evidence from multicenter randomized clinical trials has confirmed that SLNB is a safe and effective alternative surgical axillary staging approach for axillary clinical negative breast cancer with less complication and improved quality of life compared with axillary lymph node dissection (ALND). But there are specific indications and contraindictions for SLNB and special dye and multidisplinary collaboration is required. Delayed ALND should be performed if SLNB is positive, which will result in higher incidence of complication and waste of medical resource. Adequate and precise preoperative axillary staging could screen out axillary positive patients directly to receive ALND and avoid unnecessary SLNB. Axillary ultrasonography (AUS) is the most common imaging technology in axlillary staging and axillary ultrasound-guided biopsy (AUSB) is the most effective and reliable tool to avoid unnecessary SLNB with positive pathological result. Unfortunately there is lack of uniform and specified AUS diagnostic criteria and tiered standards. And there is no recommendation for the choice of AUSB. Althouth AUSB can pick up extra 25% patients directly to SLNB, there are still 25% patients suffering from reoperation with positive SLNB. In order to re-staging false negative AUSB patients, we try to build a predictive nomogram with clinicopathological and ultrasonic chataceristics, and to discover differentially expressed miRNA or protein, which could assist in preoperative axillary staging.MethodsApplication of AUS708 consecutive untreated female primary invasive breast cancer patients were prospectively studied from April 2010 to December 2014, and finally 572 patients were enrolled with record of clinicopathological and ultrasonic data. Suspicious ALN was diagnosed by fine needle biopsy (FNA) or core needle biopsy (CNB). With the gold standard of postoperative pathologic ALN status without neoadjuvant chemotherapy or positive ASUB with neoadjuvant chemotherapy, the diagnostic efficiency of three staging modalities, wihich are physical examination (PE), AUS and AUSB was compared. The diagnostic efficiency of FNA and CNB for the same lymph node was alse compared. Univariate analysis and multivariate analysis was performed to find out clinicopathological and ultrasonic factors related to ALN metastasis. Logistc regression analyses were performed to build and validate different predictive nomograms. Receiver-operator characteristic (ROC) curves were drawn to calculate areas under the ROC curves (AUC), for the purpose of evaluation of nomogram predictive efficiency. With reference to Breast Imaging Reporting and Data System (BI-RADS), ALN ultrasonic "BI-RADS" was drafted with cortical thickness as the main indicator, with recommendation of axillary staging choice.miRNA research38 RNA samples extracted from primary invasive breast cancer tissues were planned to be detected by Agilent miRNA chip and mRNA expression chip.19 ALN negative and 19 ALN positive patients were matched by age, primary tumor size and pathologic tumor type. Raw data was analyzed by Gene Spring GX software to differentiate 2 fold change expression miRNAs, which would be validated by quantitative reverse transcription polymerase chain reaction (qRT-PCR) and be used to build predictive nomogram.Serum MALDI-TOF-TOF MS researchDifferentially expressed protein peak was analysed from 190 serum samples in breast cancer patients with or without ALN metastasis by Matrix-assisted laser desorption ionization-Time of flight-Time of flight Mass Spectrometry (MALDI-TOF-TOF MS). The most significant differentially expressed protein peaks were selected as predictive biomarks with bioinformatics method. Among in the 190 patients whose serum samples were studied,172 were with invasive breast cancer (87 with ALN involved and 85 ALN uninvolved) and 18 were with non-invasive breast caner.62 ALN uninvolved and 60 ALN involved were selected as training group and validating group, the other 25 ALN uninvolved,25 ALN involved and 18 non-invasive breast caner were selected as testing group.11 patients were AUSB positive and 14 were AUSB false negative in 25 ALN involved patients in testing group. Rawa data was preprocessed and nomorlized and characteristic vectors were extracted by filtration method combined with model dependant method. Protein peak nomogram for predicting ALN metastasis was built by genetic algorithm and pattern recognition. Diagnostic efficiency of the nomogram was validated and evaluated with 10-fold cross over method in validating group and tested in testing group. Sensitivy, specificity, accuracy, negative predictive value (NPV) and positive predictive value (PPV) were calculated. Value of the nomogram in clinical application was evaluated combined with AUS in axillary staging.Results:Application of AUS572 eligible cases were studied. Sensitivy, specificity, NPV and PPV of PE was 41.8%,92.2%,82.1% and 65.1% respectively, of AUS was 77.9%,81.9% 78.5%, 81.4%, of AUSB was 69.6%,100%,100%,80%. The accuracy in axillary staing of PE, AUS, AUSB was 69.1%、80.1% and 86.3% respectively, and Youden index was 34%, 59.8%, and 69.6%. Comparison of the three modalities showed that sensitivity rank was AUS> AUSB> PE, specificity was AUSB> PE> AUS, PPV was AUSB> PE> AUS, NPV was AUS> AUSB> PE, accuracy and Youden index was AUSB> AUS> PE. There was statistically significant difference between all the groups. AUSB was performed in 214 cases, in which 151 cases FNA and 209 cases CNB. FNA and CNB were performed in 146 cases for the same targeted lymph node, and the results of FNA and CNB were compared. Sensitivy of FNA and CNB was 67.2% and 91.8%, NPVwas 37.5% and 70.6%, accuracy was 72.6% and 93.2%, specificity and PPV were both 100% in FNA and CNB. Combined diagnostic sensitivity of FNA and CNB was 95.9%, specificity was 96.6%, NPV was 82.8%. The diagnostic efficiency of FNA, CNB, FNA combined with CNB was ranked as follows, FNA combined with CNB> CNB> FNA. Two of the three methods was compared by McNemar and the result was CNB> FNA with statistically significant difference (P< 0.001), and FNA+CNB> CNB without statistically significant difference (P=0.063). There was no severe adverse effects in both of the two biopsies, which were well tolerated.572 cases were dividied into two groups,367 cases before year 2014 as training group and 205 cases in year 2014 as validating group. ALN involved incidence was 47.1% in training group and 43.9% in validating group. All the variables were balanced between training and validating group statisitically. Result of univariate and multivariate analysis showed that primary tumor size and location, ER status, PR status, Her-2 status, cortex and hilum thickness, transverse and longitude diameter of ALN, lobulation of the ALN, palpation of the ALN were significantly related to ALN metastasis. Nomogram 1 was built by logisitic regression with 28 clinicopathological and ultrasonic variables, and the predictive accuracy was 91% and 85.5% in training and validating group, AUC was 0.91 (95% CI:0.87-0.95) and 0.88 (95% CI:0.82-0.93) respectively. Nomogram 2 was built by logisitic regression with 10 ultrasonic vaviables of primary tumor and ALN, and the predictive accuracy was 84.1% and 81.8% in training and validating group, AUC was 0.93 (95% CI:0.91-0.96) and 0.91 (95% CI:0.86-0.95) respectively. Variables were decreased in nomogram 2 without decline in AUC and even minorly increased compared with nomogram 1. Although Nomogram 1 may screened out more ALN positive patients in accordance with the research objective, Nomogram 2 have more potential in clinical application with convenience.ALN ultrasonic "BI-RADS" was drafted with cortex thickness ranking. Characteristics with false negative AUSB was find out by logistic multivariate analysis, which are primary tumor size between 1-3cm, cortex thickness≤0.36cm, Grade I of WHO pathological classification and Luminal A type. With comprehensive consideration of different ALN-FNA/CNB result according to ALN ultrasonic "BI-RADS" and cost benefit analysis, recommendations was given as follows:BI-RADS 3 SLNBBI-RADS 4A SLNB/CNBBI-RADS 4B CNB/SLNBBI-RADS 4C CNB/FNABI-RADS 5 FNA/CNBmiRNA researchQuality Control of 38 RNA samples extracted from primary tissues with or without ALN metastasis was eligible and date between two chip groups was balanced with relatively low coefficient of variation. Raw data was normalized and analyzed by Gene Spring GX and there was no statistically significant differentially expressed miRNA with 2 fold changes in the two groups. If the cut-off P value is set as 0.15, miRNA-1305and miRNA-886-3p were the two differentially expressed miRNAs above 2 fold changes.4 pairs of ALN negative and positive serum samples were mixed with equal allocation as 1 pair chip, and the analysis result showed that there were series differentially expressed miRNAs such as miRNA-190a. Because of the negative result of tissue RNA chip and lack of serum sample, this part of research was temporarily terminated without the following validation and nomogram building work.Serum MALDI-TOF-TOF MS research28 protein peaks were screened out in training and validating group, with 12 peaks high expressed and 16 low expressed in ALN positive cases.15 peaks were selected to built predictive nomogram with genetic algorithm by ZJUPDAS software.7 peaks were high expressed in ALN positive cases, which are 2022,2093,2699,4055,4321,4790 and 4966.8 peaks were low expressed in ALN positive cases, which are 1597,2953, 3218,3317,5907,6435,6633 and 6650. Validating accuracy is 85.48% in ALN negative cases and 81.67% in ALN positive cases. Sensitivy, specificity, accuracy, PPV and NPV were 81.7%,85.5%,83.6%,84.5% and 82.8%. If classified 18 non-invasive breast cancer cases as ALN negative, nomogram predictive value of Sensitivy, specificity, accuracy, PPV and NPV were 24%,48.8%,39.7%,21.4% and 52.5% in testing group (P<0.05). If the 18 non-invasive breast cancer cases were kicked out, nomogram predictive value of Sensitivy, specificity, accuracy, PPV and NPV were 24%, 56%,40%,35.3% amd 42.4% in testing group (P> 0.05). If the 18 non-invasive breast cancer cases and 11 AUSB positive cases were kicked out, nomogram predictive value of Sensitivy, specificity, accuracy, PPV and NPV were 24%,56%,40%,21.4% and 56% in testing group (P> 0.05).Conclusions:Application of AUSAUSB is the most effective modality in axillary staging compared with PE and AUS. It can increase the detection rate of positive ALN cases preoperatively with avoidance of unnecessary SLNB. CNB is the preferred choice of ALN biopsy to FNA. Simplified ALN predictive nomogram with variables of primary tumor layer and calcification, cortex and hilum thickness has a favorable diagnostic efficiency with AUC 0.91, which can be validated in multicenter clinical trials and applicated in clinical use later on. ALN ultrasonic "BI-RADS" with cortex thickness ranking as its key indicator and the corresponding recommendation is a optimized preoperative axillary staing process and can lead to "patient-tailored" axillary staging.miRNA researchThe research failed to discover statistically significant differentially expressed miRNA above 2 fold changes between the ALN negative and positive cases as biomarks for ALN status prediction. It is impossible to evaluate serum miRNA expression because of too less samples. It was speculated that maybe there was no existence of sensitive biomarks of differentially expressed miRNAs in primary tumor tissue. And also maybe it was attributed to the insufficient cases in our study and should be verified with increased samples.Serum MALDI-TOF-TOF MS researchMagnetic bead method combined with MALDI-TOF-TOF MS is of low sensitivity and moderate specificity in predicting breast cancer ALN metastasis. Results of two classification of tesing group was not statistically siganificant in difference because of insufficient cases. More cases should be enrolled, especially more AUSB false negative and AUS "BI-RADS" 4 patients. Although the predictive accuracy is low in AUSB false netative cases, the nomogram can screen another extra 21.4% ALN positive patients. If it was utilized with "second-look" AUS and AUSB, benefit would be observed with more avoidance of unnecessary SLNB.