2011 SSAT PLENARY PRESENTATION

RT-PCR Increases Detection of Submicroscopic PeritonealMetastases in Gastric Cancer and Has Prognostic Significance

Joyce Wong & Kaitlyn Jane Kelly & Arjun Mittra &

Mithat Gonen & Peter Allen & Yuman Fong & Daniel Coit

Received: 18 May 2011 /Accepted: 10 February 2012 /Published online: 24 February 2012# 2012 The Society for Surgery of the Alimentary Tract

AbstractBackground Positive peritoneal cytology confers the same prognosis as clinical stage IV disease in gastric cancer. Conventionalcytology examination, however, has low sensitivity. We hypothesize that real-time polymerase chain reaction (RT-PCR) mayhave increased sensitivity and provide more accurate staging information.Methods From February 2007 to April 2009, peritoneal lavage samples were collected prospectively from 156 patients withbiopsy-proven gastric cancer undergoing staging laparoscopy. These washings were analyzed by both Papanicolaou stainingand RT-PCR for the tumor marker carcinoembryonic antigen (CEA).Results Visible peritoneal disease was seen at laparoscopy in 38 patients (LAP+, 24%). Cytology was positive (CYT+) in 23patients, while RT-PCR was positive (PCR+) in 30. The sensitivity of CYT for the detection of visible disease was 61%compared to 79% for PCR (P00.02). No visible peritoneal disease was seen at laparoscopy (LAP−) in 118 (76%) patients.Eight (7%) were CYT+, while 28 (24%) were PCR+. Predictors of PCR positivity included advanced-stage disease (T3–4 vs.T1–2 tumors) and poor pathologic features such as vascular or perineural invasion. Long-term follow-up demonstrated aworse survival of LAP−CYT−PCR+ (P00.0003) and LAP−CYT+PCR+ (P00.0004) compared to LAP−CYT−PCR−patients. There was no significant difference in survival between CYT−PCR+ and CYT+PCR+ patients. PCR positivityalso predicted a higher likelihood of disease recurrence after resection. An R0 resection was performed in 85 LAP− patients(54%): only 1 (1%) was CYT+, while 13 (15%) were PCR+. Of this group, PCR+ demonstrated a worse survival than PCR−patients (P00.02). Further analysis showed that, in R0 resection, stage III/IV, CYT− subgroup, PCR+ was associated with atrend towards worse survival (P00.09) compared to PCR− patients.Conclusion RT-PCR for CEA increases the detection of subclinical peritoneal disease and is more sensitive than cytology.Predictors of positive PCR included advanced-stage disease, vascular invasion, and perineural invasion. PCR positivity wasassociated with increased disease recurrence and decreased survival. Further follow-up is required to determine if PCRpositivity alone is an independent predictor of poor survival in gastric cancer.

Keywords Gastric cancer . Peritoneal metastases . RT-PCR Introduction

Gastric cancer remains one of the world’s most commonincident cancers, with a particularly strong prevalence inWestern Pacific countries including Japan, China, andKorea. In 2008, it was the fourth most common malignancyworldwide but was the second leading cause of cancer-related mortality.1 In the USA, the overall incidence of

J. Wong :K. J. Kelly :A. Mittra :M. Gonen : P. Allen :Y. Fong :D. Coit (*)Department of Surgery, Memorial Sloan–Kettering Cancer Center,275 York Avenue,New York, NY 10021, USAe-mail: coitd@mskcc.org

J Gastrointest Surg (2012) 16:889–896DOI 10.1007/s11605-012-1845-2

gastric cancer is declining, now representing the 14th mostcommon cancer. However, every year, approximately 21,000new gastric cancer cases are diagnosed in the USA, account-ing for over 10,000 deaths annually.2 Surgery remains themainstay of curative treatment.

Pretreatment staging typically includes endoscopy withbiopsy, imaging with multidetector computed tomography(CT), and selective use of endoscopic ultrasound. Due to thelow sensitivity of CT in the detection of peritoneal metasta-ses, however, our institution has adopted routine staginglaparoscopy to visually assess the abdominal cavity forsubradiologic metastatic disease.3 Staging laparoscopy alsoallows for the collection of peritoneal lavage fluid for cyto-logic examination with minimal morbidity.4 Positive perito-neal cytology has been shown to be an independentpredictor for disease recurrence and poor overall survival(OS) and has recently been included in the seventh editionof the American Joint Committee on Cancer (AJCC) stagingmanual as M1 disease.5–7 Peritoneal cytology is currentlyassessed using Papanicolaou staining to determine the pres-ence of malignant cells.8,9 Cytology, however, has a lowsensitivity of 54–59% and is not always positive in cases ofvisible peritoneal disease.10 Currently, patients with positivecytology go on to be treated with systemic chemotherapy andtheir peritoneal lavage reassessed after treatment. Severalstudies, including Western series, have demonstrated thatclearance of positive cytology is associated with an improveddisease-specific survival and longer OS.11,12 Clearly, evalua-tion of peritoneal cytology is paramount to stratifying patientsinto treatment algorithms, determining whether surgical resec-tion or systemic chemotherapy should be offered.

We previously evaluated the use of reverse transcriptionpolymerase chain reaction (RT-PCR) in the detection ofseveral tumor markers, including carcinoembryonic antigen(CEA), CK20, survivin, and MUC2 in peritoneal lavagefluid obtained during staging laparoscopy. Detection ofCEA mRNA proved to have the highest sensitivity andspecificity (100% and 91%, respectively) for cancer cells,consistent with other reports.13,14 While the sensitivity ofRT-PCR for the detection of free cancer cells exceeded thatof cytology, its prognostic value in the clinical treatment ofpatients remains to be determined. Several studies havesuggested that gastric cancer patients with RT-PCR CEAmRNA positive lavage fluid have a significantly worseprognosis when compared with PCR-negative patients.15,16

However, these studies have not clearly defined the prog-nostic significance of positive PCR lavage alone in thesetting of no visible peritoneal implants and negative cytol-ogy in curatively resected patients, independent of conven-tional predictors such as T and N status. Furthermore, fewstudies have been carried out in the USA or Europe dem-onstrating the value of RT-PCR analysis of peritoneal lavagefluid in gastric cancer patients.17

This study set out to determine whether RT-PCR for CEAmRNA increased the detection of nonvisible peritoneal dis-ease and whether this finding had prognostic implications ingastric cancer patients. Specifically, the aims included deter-mining the incidence of PCR+ lavage in our study population,evaluating predictors of PCR positivity, and defining theclinical significance of PCR+ lavage.

Methods

Patients

From February 2007 to April 2009, patients with biopsy-proven gastric cancer were identified preoperatively andprovided informed consent for this study. Diagnostic lapa-roscopy was performed after CT scan and endoscopy, andendoscopic ultrasound when appropriate, in order to ade-quately stage the patient prior to initial treatment. Peritoneallavage washings were collected from patients undergoingstaging diagnostic laparoscopy for both cytologic examina-tion and RT-PCR evaluation. Patients who did not havemetastatic disease, either visible or by positive cytology,were offered resection. The majority (98%) of patientswho were cytology positive by Papanicolaou stain, or hadvisible peritoneal disease, were treated with systemic che-motherapy; PCR status was not used for any treatmentdecisions. All patients were then followed up prospectivelyby interval physical exams, imaging, and clinic visits todetermine outcome. This study was approved by the insti-tutional review board.

Clinical and Pathologic Staging

All patients were evaluated prior to laparoscopy with CTscan, endoscopy, and endoscopic ultrasound when appli-cable. Patients were clinically and pathologically stagedin accordance with the AJCC sixth edition staging criteria;positive cytology, however, was considered metastaticdisease.

Laparoscopy

A three-port technique was used to perform a thoroughinspection of the abdomen, including the stomach, liver,omentum, ovaries, and the visceral and parietal peritonealsurfaces. Any suspicious visible lesions in the peritonealcavity were biopsied. Positive biopsies were classified asLAP+. Saline was then instilled during laparoscopy andlavage obtained from the right upper quadrant, left upperquadrant, and pelvis. This fluid was sent to pathology forcytopathologic examination as well as on ice to our labora-tory for RT-PCR analysis.

890 J Gastrointest Surg (2012) 16:889–896

Cytologic Examination

Specimens collected for cytology were placed in Cytolytefixative (Cytyc Corp., Marlborough, MA, USA). After cen-trifugation for 10 min, the resulting cell pellet was fixedwith PreservCyte (Cytyc Corp.). Using the Thin Prep pro-cedure, two slide preparations were created; the first wasstained with a modified hematoxylin and eosin preparation,while the second was stained using the Papanicolaou method.Patients with malignant cells visualized on microscopicexamination of these slides were classified as CYT+.

PCR Analysis

For PCR examination, three peritoneal washing samples perpatient were collected and centrifuged at 800 rpm for 5 minat 4°C. The resulting cell pellets were combined and resus-pended in 1 ml of supernatant. This was then centrifuged at8,000×g for 5 min at 4°C. The pellet was then processed toisolate RNA using the RNeasy Mini-Kit (Qiagen, Valencia,CA, USA), according to the manufacturer protocol. Briefly,the pellet was resuspended by the addition of Buffer RLTwith beta-mercaptoethanol. Samples were then homoge-nized by passage through a blunt 20-gauge needle fivetimes. Lysates were washed with 70% ethanol and trans-ferred to an RNeasy mini-column, washed with buffer RW1,and eluted in a 30-μL volume of RNase-free water. Theextracted mRNA samples were stored collectively at −80°C.

RT-PCR Controls

The gastric cancer cell line OCUM-2MD3 (Osaka CityUniversity Graduate School of Medicine, Osaka, Japan)was used as a positive control for the RT-PCR assay, as itis well known to express high levels of CEA. Cells werecultured in RPMI media and stored in a humidified incuba-tor at 37°C with 5% CO2. The endogenous control gene,18S rRNA, was used to confirm the presence of adequatemRNA in the peritoneal lavage samples.

RT-PCR

Reverse transcription was performed using the TaqManUniversal Reverse Transcriptase Master Mix (Applied Bio-systems, Foster City, CA, USA). First, the amount of RNAisolated was quantified by spectrophotometry using theNanodrop 1000 spectrophotometer (Thermo Scientific,Waltham, MA, USA). RT-PCR was performed using ran-dom hexamer priming and TaqMan Reverse TranscriptionReagents (Applied Biosystems). When the sample sizeallowed, 2 μg of RNA was used in each reaction. Whenunavailable, the entire volume of sample was used. For eachreaction, sample RNA was amplified in a 100-μL reaction

containing 20 μL deoxynucleotide triphosphate (2.5 mMeach dTNP), 22 μL MgCl2 (25 mM), 10 μL 10× RT buffer,5 μL random hexamers (50 μM), 2 μL RNase inhibitor(20 U/μL), 6.25 μL MultiScribe Reverse Transcriptase(50 U/μL), and free water (dependent on RNA volume) toa total amount of 100 μL. The samples were transferred to aThermo Hybrid thermocycler at 25°C for 10 min, 48°C for30 min, and 95°C for 5 min. The resulting cDNA sampleswere stored at −20°C.

Real-time quantitative RT-PCR was performed with theABI-PRISM 7900 HT Sequence Detection System (AppliedBiosystems) using TaqMan Assays-on-Demand GeneExpression primers for CEA and 18S rRNA (Applied Bio-systems). DNA was amplified in a 20-μL reaction contain-ing 1 μM of appropriate primer, 2 μL cDNA, and TaqManGene Expression Master Mix. Each PCR was incubated at48°C for 30 min followed by 10 min at 95°C, followed by40 cycles at 95°C for 15 s and 60°C for 60 s. Each samplewas processed over a 2-day period. Each sample was assayedin triplicate with positive PCR controls. A positive result wasdefined as any amplification of CEA mRNA in <35 cycles ofRT-PCR. These patients were classified as PCR+.

Statistical Analysis

RT-PCR was recorded as either positive or negative and corre-lated with clinical and pathologic features using a chi-squaretest. Survival curves were generated using the Kaplan–Meiermethod and compared across groups using a log-rank test.Statistical significance was defined by P<0.05.

Results

Patient Characteristics

From February 2007 to April 2009, 156 patients withbiopsy-proven gastric cancer underwent staging laparosco-py as part of their pretreatment workup. Eighty-seven (56%)patients had preoperative clinical stage I/II disease based onCT scan and endoscopy, while 69 (44%) had stage III/IVdisease. Forty-three patients (28%) had T1 or T2 tumors,while 99 (63%) had T3 or T4 tumors. Fourteen (9%)patients were listed as Tx. Slightly over one third of tumorswere located in the body and antrum each, with the remain-ing located in the cardia (N030, 19%), gastroesophagealjunction (N012, 8%), and fundus (N05, 3%).

Of 156 patients, 38 (24%) had visible peritoneal disease atlaparoscopy (LAP+); 118 (76%) were LAP−. Of the 38 LAP+patients, only 23 (61%) were CYT+; of these, all were PCR+.Of the 15 LAP+CYT− patients, 7 (47%) were PCR+. Thesensitivity of cytology for the detection of visible peritonealdisease was 61% vs. 79% for PCR (P00.02).

J Gastrointest Surg (2012) 16:889–896 891

The 15 patients who were LAP+CYT− had histologicallysimilar tumors: adenocarcinoma, either moderate or poorlydifferentiated. The majority (N013, 87%) had diffuse-typehistology. One patient had intestinal-type histology, and onepatient had mixed histology.

One hundred eighteen patients had no visible metastaticdisease at laparoscopy (LAP−). Of these patients, eight (7%)were CYT+. Of the eight LAP−CYT+ patients, seven (88%)were PCR+. Of the 110 patients who were LAP−CYT−, 21(19%) were PCR+ (Fig. 1).

Predictors of Positive PCR

The number of patients found to have visible peritoneal dis-ease increased with more advanced stage of disease. Similarly,more patients with preoperative stage III/IV disease werefound to be CYT+ and PCR+ (Table 1). Over 25% of patientswith preoperative stage III disease were found to be LAP+ andCYT+, while over 50% were PCR+. In contrast, only 1 of 31patients with preoperative stage I disease was found to be LAP+ or CYT+ and only 2 of 31 were PCR+ (Table 1). Five of 56patients with preoperative stage II disease were found to beLAP+ or CYT+, yet 17 of 56 were PCR+. Among 107patients with potentially operable preoperative stage II or IIIdisease, PCR was positive in 44 patients.

Complete pathological examination was available on 98patients who went on to resection. All 98 patients wereLAP−. Two (2%) were CYT+; 18 (18%) were PCR+. Thetwo CYT+ patients underwent resection before the cytologyresults were known. A total of 37 patients had pathologicstage 0/I disease; none of these patients was CYT+ orPCR+. Twenty-four patients had pathologic stage II disease;of these, none was CYT+ and only 1 (4%) was PCR+. Of 31pathologic stage III patients, only 1 (3%) was CYT+; how-ever, 13 (42%) were PCR+. Six patients had pathologicstage IV disease (i.e., T3N3 or T4N2 disease, based on the

2002 AJCC classification); only one (17%) was CYT+,while four (67%) were PCR+.

PCR was positive in a significant number of patients withlocally advanced disease, including patients with T3 (N016,38%) or T4 (N02, 100%) tumors, as well as N1 (N09,20%), N2 (N04, 67%), or N3 (N03, 60%) disease. Patho-logic features such as vascular invasion and perineural in-vasion have been considered poor prognostic factors. Of 52patients who demonstrated vascular invasion, 11 (21%)were PCR+. Perineural invasion was found in 47 patients;12 (26%) were PCR+. Of patients with moderately differ-entiated and poorly differentiated tumors, a significant num-ber were also PCR+, 22% and 16%, respectively (Table 2).PCR was positive in a significant proportion of patients whowere found to have poor pathologic features.

Survival and Disease Recurrence

When evaluating 110 LAP−CYT− patients who underwentresection, 15 patients recurred, 8 (53%) of whom were PCR+.Thirteen LAP−CYT− patients recurred following an R0 re-section, 7 (47%) of whom were PCR+.

At 35 months median follow-up, of the 118 LAP− patients,PCR+ patients had a worse OS compared with PCR− patients(P00.0003; Fig. 2). Among 85 patients who underwent R0resection, PCR+ patients again demonstrated a worse OScompared with PCR− patients (P00.02; Fig. 3a). Furthersubgroup analysis evaluated the outcome of R0 stage III/IV,CYT− patients relative to PCR status. PCR+ patients showeda trend towards worse OS (P0.09; Fig. 3b).

Discussion

Staging laparoscopy is important in pretreatment staging ofgastric cancer, avoiding unnecessary morbidity from

156

Yes38 (24%)

No118(76%)

Positive23 (61%)

Negative15 (39%)

Positive8 (7%)

Negative110(93%)

Positive23 (100%)

Negative0 (0%)

Positive7 (47%)

Negative8 (53%)

Positive7 (88%)

Negative1 (12%)

Positive21 (19%)

Negative89 (81%)

All patients

Visible M1 Disease

Cytology

PCR

Fig. 1 One hundred fifty-sixpatients undergoing diagnosticlaparoscopy with cytologicand RT-PCR examinationof peritoneal lavage fluid

892 J Gastrointest Surg (2012) 16:889–896

laparotomy by identifying subradiologic metastatic diseasein up to 30% of patients otherwise deemed resectable byconventional imaging. Up to 50% of patients with locallyadvanced gastric cancers may have unsuspected peritonealmetastases found at staging laparoscopy.18–20 As many as15% of patients without visible metastatic disease will have

positive peritoneal cytology by standard cytopathologicexamination.19 Patients with positive cytology have prog-nosis similar to visible stage IV disease and are unlikely tobenefit from aggressive surgical resection.5

RT-PCR has been used as a technique to evaluate thepresence of genetic biomarkers in tumor tissue and has beenable to detect micrometastatic disease associated with poorprognosis in cancers such as melanoma.21 Peritoneal lavageevaluation with RT-PCR has clinically correlated the pres-ence of amplified biomarker mRNAwith worse disease-freesurvival and OS in pancreatic and colorectal cancer.22,23

Numerous studies, particularly from Japan, have evaluatedthe clinical significance of RT-PCR in the evaluation ofperitoneal lavage fluid in gastric cancer (Table 2).14–16,24–26

These studies have demonstrated the increased sensitivity ofRT-PCR when compared to cytology for the detection ofperitoneal cancer cells in the setting of metastatic disease, afinding also confirmed in this study. PCR positivity has alsobeen significantly correlated with higher T and N stage. Thisstudy demonstrated that PCR was positive in a significantlygreater number of patients with advanced-stage disease orvascular and perineural invasion compared with those whowere CYT+. These data again suggest that PCR may be amore sensitive technique than cytology for the detection ofpotentially clinically relevant distant metastases in patientswith cancers which are more locally invasive and at a greaterrisk of recurrence.

Multiple studies of gastric cancer have shown PCR+patients to have a worse OS and earlier peritoneal recurrencethan PCR− patients (Table 3). Yonemura et al. showed thatPCR+ combined with CYT+ predicted worse OS and earlier

Table 1 Correlation between preoperative clinical stage with laparos-copy, cytology, and PCR results

Preoperative clinicalstage (N0156)

Visible peritonealdisease (LAP+)

CYT+ PCR+

I (31) 1 (3%) 1 (3%) 2 (6%)

II (56) 5 (9%) 5 (9%) 17 (30%)

III (51) 15 (29%) 15 (29%) 27 (53%)

IV (18) 17 (94%) 9 (50%) 12 (67%)

Table 2 Clinicopathologic features of 98 patients undergoing resec-tion for gastric cancer

Pathologic features (N098) CYT+ PCR+

T stage

T0/T1 (20) 0 0

T2 (34) 0 0

T3 (42) 2 (5%) 16 (38%)

T4 (2) 0 2 (100%)

N stage

N0 (42) 0 2 (5%)

N1 (45) 0 9 (20%)

N2 (6) 1 (17%) 4 (67%)

N3 (5) 1 (20%) 3 (60%)

AJCC stage

0–I (37) 0 0

II (24) 0 1 (4%)

III (31) 1 (3%) 13 (42%)

IV (6) 1 (17%) 4 (67%)

Pathologic features

Vascular invasion (52) 2 (4%) 11 (21%)

Perineural invasion (47) 2 (4%) 12 (26%)

Differentiation

Well (5) 0 0

Moderately (41) 1 (2%) 9 (22%)

Poor (49) 1 (2%) 8 (16%)

Undifferentiated (1) 0 0

Histology

Intestinal-type (42) 1 (2%) 7 (17%)

Mixed (16) 1(6%) 3 (19%)

Diffuse-type (23) 0 3 (13%)

A greater proportion of patients with advanced T and N stage and patho-logic AJCC stage disease were PCR+ compared with cytology. Addition-ally, more patients with moderately to poorly differentiated tumors,vascular invasion, and perineural invasion were PCR+ than CYT+

Follow up (months)

Sur

viva

l Dis

trib

utio

n

___ PCRPCR+

P=0.0003

N=90

N=28

Fig. 2 LAP− patients undergoing resection, PCR+ (dashed line, N028)vs. PCR− (solid line,N090). PCR+ patients had a significantly worse OScompared with PCR− patients

J Gastrointest Surg (2012) 16:889–896 893

disease recurrence. In their study, CYT−PCR− had significant-ly better OS compared to CYT+, CYT−PCR+, and CYT+PCR+ groups, which did not significantly differ from oneanother. Kodera and Wang both demonstrated that PCR+patients had significantly worse OS and disease-free survivalthan PCR− patients. Kodera found that PCR+CYT− patientshad significantly worse outcomes than CYT− patients. None ofthese studies, however, stratified patient groups by T and Nstatus or by disease stage. As the presence of PCR+ correlatesstrongly with stage of disease, the significance of positive RT-PCR, independent of cytology, tumor size, and nodal status,remains undefined.

This study reports the incidence, predictors, and clinicalsignificance of RT-PCR for CEA mRNA in peritoneal lavage

fluid of gastric cancer patients treated at a single Westerninstitution. We found that the OS of all patients who wereLAP−PCR+ was significantly worse than LAP−PCR−patients. Of those LAP− patients who underwent an R0 resec-tion, PCR+ patients again had a significantly worse OS andmore frequently developed recurrent disease compared withPCR− patients. In an attempt to determine whether PCR statuswas a predictor of survival independent of other prognosticfactors, CYT−, resected patients with negative margins, stageIII/IV were evaluated. PCR+ patients demonstrated a trendtowards worse OS, although this was not statistically signifi-cant. We did not routinely obtain preoperative serum CEAlevels for every patient in this study. As we continue to follow-up this cohort and enroll patients prospectively, however, we

N=72

N=13

N=17

N=12

Fig. 3 Survival curves ofR0-resected patients (a) and R0,stage III/IV, CYT− patients(b) by PCR status. Among 85patients who underwent R0resection, PCR+ (N013)patients had a significantlyworse survival than PCR−(N072) patients (P00.02).Among the 29 R0-resected,stage III/IV, cytology-negativepatients, PCR+ (N012) had atrend towards worse survivalthan PCR− (N017) patients(P00.09)

Table 3 Representative studies investigating the utility of RT-PCR evaluation for CEA in peritoneal fluid from gastric cancer patients

Year Author Country ofstudy

Number ofpatients

Percent PCRpositive

Clinical association and comparison to cytology

2001 Yonemura Japan 230 17 Combination PCR+ and CYT+ higher sensitivity in predictingdisease recurrence; CYT−PCR+ had similar OS to CYT+

2002 Kodera Japan 189 29 CYT+PCR+ and CYT−PCR+ worse OS compared with PCR−;no comparison with predictors such as T or N status

2004 Oyama Japan 195 28 PCR+ worse OS and earlier disease recurrence; greaterassociation with T3 and T4 disease; no comparison withcytology

2005 Wang Taiwan 40 28 PCR+ higher peritoneal recurrence rate; more sensitive thanCYT but no comparison of CYT−PCR+ patients

2006 Kodera Japan 274 36 PCR+ worse OS and earlier disease recurrence with increasedassociation with T3 and T4 disease; no comparison to cytology

2007 Katsuragi Japan 116 40 T3 and T4 or stage III/IV PCR+ patients had worse OS; nocomparison with cytology

2011 Current study USA 156 37 PCR+CYT− patients had significantly worse OS in patients whounderwent R0 resection and showed a trend towards worsesurvival in stage III/IV R0-resected CYT− patients

894 J Gastrointest Surg (2012) 16:889–896

may be able to correlate serum CEA levels with RT-PCRresults.

This study focused on the prognostic yield of utilizingRT-PCR for the detection of a single tumor marker (CEA)mRNA in peritoneal lavage as a possible indicator of clin-ically relevant disease, and therefore, a useful predictor ofadverse outcome. It is possible that the prognostic yieldcould be increased by testing for a panel of tumor markers.Future studies may include performing RT-PCR on multiplebiomarkers and analyzing prognosis based on the “genesignature” of the peritoneal lavage. With high-output andsingle-step RT-PCR kits becoming more readily available, ifPCR of peritoneal lavage fluid can be defined as a useful testin the management of gastric cancer patients, the technologymay soon be in place to enable ready access in clinicalpractice. A gene signature may differ from one tumor to thenext and may need to be defined by genetic analysis ofpretreatment biopsy. However, it is evident that, with limitedfollow-up in a small number of patients, positive PCR forCEA does not appear to be as clinically prognostic as positivecytology. Any increased yield by multimarker assay will needto be carefully evaluated for independent prognostic relevancein the setting of other known prognostic variables.

Conclusion

RT-PCR for CEA mRNA increases the detection of submi-croscopic peritoneal disease and is more sensitive than con-ventional cytology. PCR was positive in a greater proportionof patients with poor pathologic features, including advancedT and N stage and perineural and vascular invasion. Themajority of patients who recurred after resection were PCR+. PCR positivity was significantly associated with decreasedsurvival in the overall cohort as well as patients who under-went R0 resection. Longer follow-up with more patients isrequired to determine if PCR of peritoneal lavage fluid can beused both as an independent predictor of poor survival ingastric cancer and to help select patients who are unlikely tobenefit from surgical resection. Future studies will also helpdetermine whether analysis of multiple tumor markers ratherthan a single gene may increase the diagnostic yield andindependent predictive value of RT-PCR.

References

1. Ferlay J, Shin HR, Bray F, Forman D, Mathers C and Parkin DM.GLOBOCAN 2008, Cancer Incidence and Mortality Worldwide:IARC CancerBase No. 10 [Internet]. Lyon, France: InternationalAgency for Research on Cancer; 2010. Available from: http://globocan.iarc.fr

2. American Cancer Society.: Cancer Facts and Figures 2010.Atlanta, GA: American Cancer Society, 2010

3. Kim SJ, Kim HH, Kim YH, et al. Peritoneal metastasis: detectionwith 16− or 64-detector row CT in patients undergoing surgery forgastric cancer. Radiology, 2009; 253(2): 407–15.

4. Muntean V, Mihailov A, Iancu C, et al. Staging laparoscopy ingastric cancer. Accuracy and impact on therapy. Journal of Gas-trointestinal and Liver Diseases, 2009; 18(2): 189–95.

5. Bentrem D, Wilton A, Mazumdar M, et al. The value of peritonealcytology as a preoperative predictor in patients with gastric carci-noma undergoing curative resection. Annals of Surgical Oncology,2004; 12(5): 1–7

6. La Torre M, Giovagnoli MR, Sforza N, et al. Peritoneal washcytology in gastric carcinoma. Prognostic significance and thera-peutic consequences. European Journal of Surgical Oncology,2010; Oct; 36(10): 982–6.

7. Edge SB, Byrd DR, Compton CC, et al. American Joint Commit-tee on Cancer (AJCC) Cancer Staging Manual (7th ed.). Chicago:Springer Inc., 2010.

8. Nakajiama T, Hirashima S, Hirata M, et al. Prognostic and thera-peutic values of peritoneal cytology in gastric cancer. Acta Cytol,1978; 22: 225–229.

9. Kodera Y, Yamamura Y, Shimizu Y, et al. Peritoneal WashingCytology: Prognostic Value of Positive Findings in Patients withGastric Carcinoma Undergoing a Potentially Curative Resection.Journal of Surgical Oncology, 1999; 72: 60–65.

10. Wilkiemeyer MB, Bieligk SC, Ashfaq R, et al. Laparoscopy aloneis superior to peritoneal cytology in staging gastric and esophagealcarcinoma. Surgical Endoscopy, 2004; 18(5): 852–856.

11. Lorenzen S, Panzram B, Rosenberg R, et al. Prognostic signifi-cance of free peritoneal tumor cells in the peritoneal cavity beforeand after neoadjuvant chemotherapy in patients with gastric carci-noma undergoing potentially curative resection. Annals of SurgicalOncology, 2010; 17(10): 2733–9.

12. Mezhir JJ, ShahMA, Jacks LM, et al. Positive Peritoneal Cytology inPatients with Gastric Cancer: Natural History and Outcome of 291Patients. Annals of Surgical Oncology, 2010; 17(12): 3173–80.

13. Moore Dalal K, Woo Y, Kelly K, et al. Detection of micrometa-stases in peritoneal washings of gastric cancer patients by thereverse transcriptase polymerase chain reaction. Gastric Cancer,2008; 11:206–213.

14. Kodera Y, Nakanishi H, Ito S, et al. Quantitative detection ofdisseminated free cancer cells in peritoneal washes with real-timereverse transcriptase-polymerase chain reaction: a sensitive predic-tor of outcome for patients with gastric carcinoma. Annals ofSurgery, 2002; 235: 499–506.

15. Katsuragi K, Yashiro M, Sawada T, et al. Prognostic impact ofPCR-based identification of isolated tumour cells in the peritoneallavage fluid of gastric cancer patients who underwent a curative R0resection. British Journal of Cancer, 2007; 97: 550–556.

16. Yonemura Y, Endou Y, Fujimura T, et al. Diagnostic value ofpreoperative RT-PCR-based screening method to detect carcinoem-bryonic antigen-expressing free cancer cells in the peritoneal cavityfrom patients with gastric cancer. ANZ J Surgery, 2001; 71: 521–8.

17. Fujiwara Y, Doki Y, Taniguchi H, et al. Genetic detection of free cancercells in the peritoneal cavity of the patient with gastric cancer: presentstatus and future perspectives. Gastric Cancer, 2007; 10: 197–204.

18. Nath J, Moorthy K, Taniere P, et al. Peritoneal lavage cytology inpatients with oesophagogastric adenocarcinoma. British Journal ofSurgery, 2008; 95: 721–726.

19. Hur H, Lee HH, Jung H, et al. Predicting Factors of UnexplainedPeritoneal Seeding in Locally Advanced Gastric Cancer: Indica-tions for Staging Laparoscopy. Journal of Surgical Oncology,2010; 102: 753–757.

20. Song KY, Kim JJ, Kim SN, et al. Staging Laparoscopy for AdvancedGastric Cancer: Is It also Useful for the Group Which Has anAggressive Surgical Strategy? World Journal of Surgery, 2007; 31:1228–1233.

J Gastrointest Surg (2012) 16:889–896 895

21. Nicholl MB, Elashoff D, Takeuchi H, et al. Molecular upstagingbased on paraffin-embedded sentinel lymph nodes: ten-year follow-up confirms prognostic utility in melanoma patients. Annals ofSurgery, 2011; 253(1): 116−22.

22. Kelly KJ, Wong J, Gladdy R, et al. Prognostic Impact of RT-PCR-Based Detection of Peritoneal Micrometastases in Patients with Pan-creatic Cancer Undergoing Curative Resection. Annals of SurgicalOncology, 2009; 16: 3333–3339.

23. Lloyd JM, McIver CM, Stephenson SA, et al. Identification ofEarly-Stage Colorectal Cancer Patients at Risk of Relapse Post-Resection by Immunobead Reverse Transcription-PCR Analysis ofPeritoneal Lavage Fluid for Malignant Cells. Clinical Cancer Re-search, 2006; 12: 417–423.

24. Oyama K, Terashima M, Takagane A, et al. Prognostic signifi-cance of peritoneal minimal residual disease in gastric cancerdetected by reverse transcription-polymerase chain reaction. BritishJournal of Surgery, 2004; 91(4): 435–43.

25. Wang JY, Lin SR, Lu CY, et al. Gastric cancer cell detection inperitoneal lavage: RT-PCR for carcinoembryonic antigen tran-scripts versus the combined cytology with peritoneal carcinoem-bryonic antigen levels. Cancer Letters, 2005; 223(1): 129–35.

26. Kodera Y, Nakanishi H, Ito S, et al. Prognostic Significance ofIntraperitoneal Cancer Cells in Gastric Carcinoma: Analysis of RealTime Reverse Transcriptase-Polymerase Chain Reaction after 5Years of Followup. Journal of the American College of Surgeons,2006; 202(2): 231–236.

Discussant

Dr. R. Daniel Beauchamp (Nashville, TN, USA): One limitation ofperitoneal cytology for staging of gastric cancer is that is has a rela-tively low sensitivity, and may be negative, even in the presence ofvisible peritoneal disease. This paper by Joyce Wong, Dan Coit, andcolleagues from Memorial Sloan–Kettering Cancer Center examinesthe use of quantitative real-time RT-PCR analysis of the tumor markercarcinoembryonic antigen (CEA) as a more sensitive method to detectevidence of occult metastatic spread of within the peritoneal cavity in a

group of patients with gastric adenocarcinoma. In addition, the pres-ence of PCR positivity was associated with significantly worse survivalas compared with PCR-negative patients, although PCR-positive statusdid not achieve statistical significance in cytologically negativepatients. Interestingly, 39% of patients who had laparoscopically de-tectable peritoneal disease were cytologically negative and, of these,only 47% were PCR positive. Thus, my first two questions are thefollowing: Did these PCR-negative tumors express CEA, and is thereperhaps a more sensitive and specific gastric cancer cell biomarker orset of biomarkers that may be used? Secondly, since the peritonealwashings were done by laparoscopy, using this information for deci-sions to proceed with resection in the absence of visible implantswould require awaiting these results and a second procedure. Havethe authors begun to use this approach for decisions regarding resec-tion, and can it be done with peritoneal lavage under local anesthesia?

Closing Discussant

Dr. Joyce Wong: Thank you for the discussion of our paper. Address-ing your first question, we did also note that PCR for CEA was notpositive in all or the vast majority of patients with positive cytology.While we do not have all the pathology data pertaining to CEAspecifically, we presume that not all gastric cancers are CEA-expressing. We did examine a number of biomarkers prior to initiatingthis study, including CK20 and several of the MUC genes; however,CEAwas the single most sensitive marker. We chose to independentlyexamine RT-PCR for CEA, although it would be useful in the future todetermine whether a panel of biomarkers may enhance disease de-tection. As to your second point, we performed staging laparoscopyas a separate procedure so that cytology results were available atthe time of resection. Our institution considers patients with posi-tive cytology as unresectable and will stratify those patients to-wards chemotherapy. Laparoscopy was performed with minimalmorbidity. There has been some discussion revolving around howto perform peritoneal lavage without general anesthesia, althoughwe currently do not routinely perform any other approach forobtaining peritoneal lavage.

896 J Gastrointest Surg (2012) 16:889–896