发布了文章
MultiplePrimaryMalignancie
Multiple Primary Malignancies in Patients With Hepatocellular CarcinomaA Largest Series With 26-Year Follow-UpWei Xu, MD, Yilei Mao, MD, PhDMedicine _ Volume 95, Number 17, April 2016北京协和医院肝脏外科毛一雷Abstract: Multiple primary malignancies (MPMs) are defined as 2 ormore malignancies without subordinate relationship detected indifferent organs of an individual patient. Reports addressing MPMpatients with hepatocellular carcinoma (HCC) are rare. We perform a26-year follow-up study to investigate characteristics and prognosis ofMPM patients associated with HCC due to the scarcity of relativeresearches.We retrospectively analyzed records of 40 patients who werediagnosed with MPM including HCC at the Departments of Surgeryat Peking Union Medical College Hospital during 1989 to 2010. Theirclinical characteristics and postoperative survival were compared withthose of 448 patients who had HCC only during the study period.Among the 40 MPM patients, 11 were diagnosed synchronously and29 metachronously. The most common extra-hepatic malignancies werelung cancer (15%), colorectal (12.5%), and thyroid carcinoma (12.5%).MPM patients had a negative hepatitis B virus infection rate (P¼0.013)and lower median alfa-fetoprotein (AFP) level (P¼0.001). Post-operative1-, 3-, and 5-year overall survival (OS) rates for MPM patients were82.5%, 64.5%, and 38.6% respectively, and showed no significant differencewith those of HCC-only patients (84.7%, 54.2%, and 38.3%P¼0.726). During follow-up, 24 MPM patients died, including 17(70.8%) who died of HCC-related causes. In univariate analysis, synchronousdiagnosis, higher gamma glutamyltransferase (GGT) and/orAFP levels, tumor >5 cm and vascular invasion were significantlyassociated with shorter OS, but only tumor size was an independentOS factor in Cox modeling analysis.HCC should be considered as a potential second primary for allcancer survivors. Most MPM patients died of HCC-related causes andshowed no significant difference in OS compared with HCC-onlypatients. Tumor size of HCC, rather than MPMs itself, was the onlyindependent OS predictor for the MPM patients.(Medicine 95(17):e3491)Abbreviations: AFP = alfa-fetoprotein, ALT = alaninetransaminase, CI = confidence interval, GGT = gammaglutamyltransferase, HBsAg = hepatitis B virus surface antigen,HBV = hepatitis B virus, HCC = hepatocellular carcinoma, HCV =hepatitis C virus, HR = hazard ratio, MPMs = multiple primarymalignancies, OS = overall survival.INTRODUCTIONMultiple primary malignancies (MPMs) were firstdescribed according to the 1932 definition of Warrenand Gates: each tumor has to present definite attributes ofmalignancy, the tumors have to be histological distinctiveand the possibility of one being a metastasis of the other mustbe ruled out.1 Thanks to continually improving screeningprograms, diagnostic, and treatment methods, survival ratesfor newly diagnosed cancer patients are increasing. Thisimprovement has led to a steady increase in the number of newlydiagnosedMPMpatients.2 In the United States,MPMs constitute18% of all cancers diagnosed; in European countries, such as theCzech Republic, the MPM incidence is more than 11%.3Hepatocellular carcinoma (HCC) ranks fifth in cancerincidence and third in cancer mortality worldwide.4 Althoughless than 1% of MPM patients reported had HCC in 1990s,5longer overall survival (OS) of oncology patients elevated therisk of MPM significantly. By 2002, liver cancer was frequentlydiagnosed with other major malignant tumors; it was found in11.5% of all MPMs in Korea.6 MPM patients who develop HCCover a long-term follow-up are no longer considered unusual,and clinicians increasingly need to consider the development ofmultiple primary cancers with HCC.Information regarding the MPM patients with HCC isimportant, as it could clarify etiological factors and may verifythe need to screen for associated malignancies during patientfollow-up. Understanding of clinicopathological features andprognostic factors are also needed to facilitate appropriatemanagement of MPM patients. However, knowledge of characteristicand outcomes of MPM patients remains limited.Editor: Zhentian Li.Received: December 1, 2015; revised: March 16, 2016; accepted: March31, 2016.From the Department of Liver Surgery, Peking Union Medical CollegeHospital, Chinese Academy of Medical Sciences and Peking UnionMedical College, 1# Shuai-Fu-Yuan, Wang-Fu-Jing, Beijing, 100730,China.Correspondence: Yilei Mao, Department of Liver Surgery, Peking UnionMedical College Hospital, Chinese Academy of Medical Sciences andPeking Union Medical College, 1# Shuai-Fu-Yuan, Wang-Fu-Jing,Beijing, 100730, China (e-mail: pumch-liver@hotmail.com).The 9 authors are justifiably credited with authorship, according to theauthorship criteria. In detail: study concept and design: Wei Xu, YileiMao; Acquisition of data:Wei Xu,Wenjun Liao, Penglei Ge, Jinjun Ren,Haifeng Xu, Huayu Yang; Analysis and interpretation of data: Wei Xu,Haifeng Xu, Huayu Yang; Drafting of the manuscript: Wei Xu, YileiMao; Critical revision of the manuscript for important intellectualcontent: Wei Xu, Xinting Sang, Xin Lu, Yilei Mao; Statistical analysis:Wei Xu, Haifeng Xu, Huayu Yang; Technical or material support:Xinting Sang, Xin Lu; Study supervision: Wei Xu, Huayu Yang, XintingSang, Xin Lu; Final approval: Yilei Mao.This study was supported by National Key Technology Research andDevelopment Program of China 2012 (grant number BAI06B01),National Natural Science Foundation of China (81201566), the NationalHigh Technology Research and Development Program (‘‘863’’ Program)of China (2015AA020303), and the Specialized Research Fund for theDoctoral Program of Higher Education (20121106110002).The authors declare no conflict of interest.Copyright # 2016 Wolters Kluwer Health, Inc. All rights reserved.This is an open access article distributed under the Creative CommonsAttribution License 4.0, which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.ISSN: 0025-7974DOI: 10.1097/MD.0000000000003491Medicine®OBSERVATIONAL STUDYMedicine _ Volume 95, Number 17, April 2016 www.md-journal.com | 1To our knowledge, only studies with cohorts of 30 patientsor fewer have been performed in Japan or Western countries forMPM patients with HCC who had received radical hepatectomy.7,8 The clinicopathologic characteristics and outcomes ofMPM patients are poorly understood, especially in Asiancountries. This retrospective study includes the largest samplesize than any other researches and 26 years follow-up time,in order to characterize MPM patients and to explore theirlong-term prognosis.METHODSBetween January 1989 and September 2010, 40 patientswith HCC that had been treated with radical hepatectomies werediagnosed with extra-hepatic primary malignancies at ourinstitution; we regarded these patients as the target group(MPM group). Over the same period, 448 others with HCConly received hepatectomies; these patients were defined as thecontrol group. In both groups, HCC was diagnosed on the basisof the histopathology from hepatectomy samples. The extrahepaticprimary malignancies were diagnosed on the basis ofhistopathology from resection (36/40) or biopsy (4/40) samples.A diagnosis of HCC as a second primary malignancy should bepathologically confirmed, as the liver is a common site formetastases and imaging findings may be atypical. To avoid thepossibility of misdiagnosis between HCC and metastatic carcinoma,MPM patients who were diagnosed only by clinicalmethods were not included in this analysis. The MPM groupwas further classified into synchronous (2 malignancies diagnosedwithin a 6-month period) or metachronous (detected morethan 6 months apart). The study protocol was approved by theEthics Committee of Peking Union Medical College Hospital.The preoperative data of patients’ clinical characteristicsincluding age, sex, family history, serum hepatitis B virus(HBV), surface antigen (HBsAg), hepatitis C virus (HCV) antibody,serum alfa-fetoprotein (AFP) were collected, and histopathologicinformation regarding tumor number and size, tumorlocation, vascular invasion, nodal status, and cirrhotic change inbackground liverwere recorded.Tumordifferentiationwas gradedby theEdmondson grading system.9TheTNMstaging systemwasused to assessHCCstage.Time forHCCsurgeries, blood loss, andbloodtransfusionwere recorded.Time for surgerieswas definedasthe time fromthe beginning of surgery topatients’ awakeningfromanesthesia. Median survival, and cumulative 3-year and 5-yearsurvival rates were calculated. OS was defined as the intervalbetween surgery and death or the last date of follow-up. Curativetherapy for the extra-hepatic primarymalignancieswas defined astreatmentwithintent tocure, suchas the surgeries formalignanciesof breast, thyroid, digestive system, and respiratory system, orradial or chemical therapy formalignancies of blood system,whileother treatment methods were regarded as palliative therapy.Clinical and pathological factors were compared usingeither Fisher exact test or Pearson x2-test, as appropriate. Thesurvival rate was calculated using the Kaplan–Meier method.COX-regression analysis was performed to identify independentrisk factors with hazard ratio (HR) and 95% confidence interval(CI). P<< span="">0.05 was considered statistically significant. Dataanalysis was performed using SPSS 19.0 software.RESULTSPatient CharacteristicsOf the 40 MPMpatients, 11 were diagnosed synchronously,and 29 metachronously, with HCC; 18 patients’ extra-hepaticprimary malignancies occurred prior to their HCC diagnoses(prior group), and 11 after their HCC diagnoses (post group). Themost sites preceding or following HCC diagnoses were lung (6/40, 15%), colorectal (5/40, 12.5%), thyroid (5/40, 12.5%), breast(3/40, 7.5%), prostate (3/40, 7.5%), and sensory organs (3/40,7.5%); 26 patients were treated by curative therapy and 4 bypalliative therapy (Table 1).Although diagnostic intervals between the 2 cancers rangedfrom 10 months to 21 years in the metachronous group(68.17_73.99 months), 51.7% (15/29) of the metachronouspatients were diagnosed with secondary cancers within 3 yearsof the initial cancer diagnosis (Figure 1). Moreover, 27.6% (8/29) of theMPMgroup were diagnosed after more than 6 years—all in the prior group, whose median interval time was significantlylonger than that of the post group (93.89_84.26 monthsvs. 26.09_10.98 months, P¼0.003).The MPM group included 36 men and 4 women. Wedetected HBsAg in 57.5% (23/40) patients; HCV antibodywas positive in 17.5% (7/40); cirrhosis was present in 62.5%(25/40). Interestingly, we found that the proportion of patientswith larger tumors (diameter >5 cm) in the synchronous groupwas significantly higher than that in metachronous group (9/11vs.12/29 P¼0.034; Table 2). No other significant differenceswere found between the synchronous and metachronous groups.Compared clinicopathological features between MPMpatients and HCC patients in control group are shown inTable 3. The mean age of diagnosis in the MPM group wassignificantly older than that in the control group (62.58_11.32TABLE 1. Site Distribution of Extra-Hepatic Primary Malignanciesin Patients With HCCMetachronousGroupLocationSynchronousGroupPriorGroupPostGroup TotalDigestive system 4 (10%) 4 (10%) 4 (10%) 12 (30%)Esophagus 0 1C 1C 2Stomach 0 2C 0 2Small intestine 1C 0 0 1Colorectal 2C 1C 2C 5Gall bladder 1C 0 1C 2Head and neck 2 (5%) 4 (10%) 4 (10%) 10 (25%)Thyroid 0 3C 2C 5Sensory organ 1C 0 2C 3Vocal cord 1C 1C 0 2Respiratorysystem2 (5%) 2 (5%) 2 (5%) 6 (15%)Lung 2C 2C 21C1P 6Urinary system 2 (5%) 2 (5%) 1 (2.5%) 5 (12.5%)Prostate 0 2P 1P 3Ureter 2C 0 0 2Breast 0 3C (7.5%) 0 3 (7.5%)Skin 0 2C (5%) 0 2 (5%)Nervous system 0 1C (2.5%) 0 1 (2.5%)Blood system 1C (2.5%) 0 0 1 (2.5%)Total 11 18 11 40Treatments for extra-hepatic primary malignancies include curativetherapy (C) and palliative therapy (P). HCC¼hepatocellular carcinoma.Xu et al Medicine _ Volume 95, Number 17, April 20162 | www.md-journal.comCopyright # 2016 Wolters Kluwer Health, Inc. All rights reserved.years vs. 55.69_11.73 years, P<< span="">0.001). Although more thanhalf of MPM patients’ HBsAg statuses were positive (57.5%),the proportion of patients in positive HBsAg status in controlgroup was significantly higher (76.3%) (P¼0.013). Further,more patients in control group showed abnormal serum AFPlevel (P¼0.001). However, no pathological features showedsignificant differences between the 2 groups.Surgical ProceduresAll MPM patients underwent surgeries for HCC includingradical liver resections, as bi- or double segmentomies (n¼19),single segmentomies (n¼7), left lateral sectorectomies (n¼4),right anterior sector-plus segmentomies (n¼3), right anteriorsectorectomies (n¼2), right hepatectomies (n¼2), rightposterior sectorectomy (n¼1), left hepatectomy (n¼1), andleft hepatectomy plus segmentomy (n¼1). Simultaneously, 4patients underwent removal of portal vein tumor thrombi, 4 hadextra-hepatic primary malignancies resected; 2 underwent cardiacperipheral vascular disconnections; and 3 received lymphnode dissections because of enlarged nodes in the hepatoduodenalligament region, including 2 found by intraoperativeexploration and 1 whose suspected lymph node metastasiswas diagnosed by preoperative magnetic resonance imaging(MRI). In 17 patients, we used Pringle’s maneuver for intermittenthepatic inflow occlusion during surgery. Mediansurgery time was 180 min (range: 100–420 min) and medianblood loss was 225mL (range: 100–2000 mL). No patients diedin the perioperative period.Surgeries for the HCC-only patients included radical liverresections as bi- or double segmentomies (n¼143), single segmentomies(n¼93), right anterior sectorectomies (n¼59), rightposterior sectorectomies (n¼42), left lateral sectorectomies(n¼25), right hepatectomies (n¼24), left hepatectomies(n¼24), right anterior sector-plus segmentomies (n¼16), lefthalf liver sector-plus segmentomies (n¼9), right half liversector-plus segmentomies (n¼8), and right half liver plus leftlateral sectorectomies (n¼5). Fourteen patients underwentremoval of portal vein or inferior vena cava tumor thrombus,FIGURE 1. Diagnosis of secondary cancer by follow-up time afterdiagnosis of the first primary tumor, among patients whose firstcancers were HCC (post), whose secondary cancers were HCC(prior), and those whose cancers were discovered more than 6months apart (metachronous)._The post and prior groups differedsignificantly at interval time >72 months (P<< span="">0.05). HCC ¼hepatocellular carcinoma.TABLE 2. Comparison of Clinicopathological CharacteristicsBetween Patients With Synchronous Group and MetachronousDiagnosesCharacteristicsSynchronousGroupMetachronousGroup P_Age, yy 0.173_62.5 (n¼18) 7 11>62.5 (n¼22) 4 18Sex 0.560Male (n¼36) 11 25Female (n¼4) 0 4HBsAg status 0.079Negative (n¼17) 2 15Positive (n¼23) 9 14HCV antibody 0.159Negative (n¼33) 11 22Positive (n¼7) 0 7ALT, U/L 1.000_40 (n¼20) 6 14>40 (n¼20) 5 15GGT, U/L 0.715_67 (n¼26) 8 18>67 (n¼14) 3 11AFP, ng/mL 0.147_20 (n¼26) 5 21>20 (n¼14) 6 8CA 19–9, U/mL 0.182_37 (n¼32) 7 25>37 (n¼8) 4 4Tumor size, cm 0.034_5 (n¼19) 2 17>5 (n¼21) 9 12Tumor location 1.000Right liver (n¼29) 8 21Left liver (n¼7) 2 5Both (n¼4) 1 3Multiple tumors 0.319No (n¼34) 8 26Yes (n¼6) 3 3Vascular invasion 0.298No (n¼35) 11 24Yes (n¼5) 0 5Edmondson grade 0.728I–II (n¼24) 6 18III–IV (n¼16) 5 11Cirrhosis 0.158No (n¼15) 2 13Yes (n¼25) 9 16Nodal status 1.000Negative (n¼39) 11 28Positive (n¼1) 0 1TNM staging 1.000I–II (n¼36) 10 26III–IV (n¼4) 1 3‘‘Bold’’ value means the ‘‘P’’ value is less than 0.05.AFP¼alfa-fetoprotein, ALT¼alanine transaminase, GGT¼gammaglutamyltransferase, HBsAg¼hepatitis B virus surface antigen,HCV¼hepatitis C virus._Fisher exact test or Pearson x2-test.yPatients’ age was divided by the median age.Medicine _ Volume 95, Number 17, April 2016 Multiple Primary Malignancies Associated With Hepatocellular CarcinomaCopyright # 2016 Wolters Kluwer Health, Inc. All rights reserved. www.md-journal.com | 36 patients underwent splenectomy and cardiac peripheral vasculardisconnection, 2 patients underwent phemister surgerysimultaneously, and 200 patients underwent inflow vascularocclusion using Pringle’s maneuver as mentioned above. Mediantime for surgery was 200 min (range: 60–600 min) and medianblood loss was 400mL (range: 50–15,000 mL). Four patientsdied in the perioperative period.The 2 groups did not significantlydiffer in surgery time (P¼0.099) or blood loss (P¼0.130).Patient PrognosisMedian follow-up time after HCC surgeries was 41.5months (range: 2 months to 8.2 years). During the follow-up,13 (32.5%) patients were still alive, 17 (42.5%) patients died ofHCC-related causes, 2 (5%) of extra-hepatic primary malignancies-related causes and 5 (12.5%) of unclear causes. Three(7.5%) patients were unconnected for various reasons. Postoperative1-, 3-, and 5-year survival rates for the 40 MPMpatients were 82.5%, 64.5%, and 38.6%, respectively.The effects of clinicopathological characteristics on survivalwere evaluated. Synchronous diagnosis, higher levels ofGGT and AFP, tumor diameter >5 cm, and vascular invasionwere significantly associated with poorer OS in univariateanalysis (Table 4), but in Cox-multivariate analysis, only tumorsize remained an independent predictor of survival (Table 5).The impact of second primary tumor on HCC survival wasalso estimated. Post-operative 1-, 3-, and 5-year survival ratesfor 448 HCC-only patients were 84.7%, 54.2%, and 38.3%,respectively, and did not significantly differ from those of theMPM group (P¼0.726, Figure 2C).DISCUSSIONPatients with malignancies have received increasing survivalbenefits from continuous progress in early cancer detection,diagnostic sub-classification, and targeted treatments.Along with increased life expectancy, cancer survivors are athigher risk of developing another malignancy compared withthe general population. Reportedly, the prevalence of MPMshas increased, and 11.0% to 21.0% of all cancers have more thanone primary in Western countries.10 The Surveillance, Epidemiologyand End Results Program of the US National CancerInstitute estimated that 7.9% of cancer survivors were livingwith a history of more than 1 primary malignancy and MPMsnow account for 16% of the newly diagnosed malignancies.11Further, any survivor of cancer has twice the probability ofdeveloping a new second primary cancer than a cancer-freeindividual of the same age and sex.12 Thus, an increasing needexists to determine subsequent cancer risks, and to provideappropriate surveillance and management. Case reports orsmall-sized studies of MPMs that include HCC have beenpublished in recent years,13–15 but information about theircharacteristics and outcomes is still limited, especially for thosewho underwent surgeries for HCC. In our series we had 40MPM patients, the largest sample size ever, receiving radicalresections for HCC and were diagnosed basis on their histopathology.Although the etiology of HCC in MPM patients remainsunclear, some evidence may be provided by their clinicalfeatures. HCC commonly arises in a background of chronichepatitis and cirrhosis in Asian countries.16 In our study, 57.5%MPM patients had positive HBsAg statuses, which was significantlyless than that the HCC-only control group (76.3%). HCVinfection has also been suggested as a potential risk factor forHCC. Our study showed that 17.5% patients in MPM groupTABLE 3. Compared Clinicopathological CharacteristicsBetween MPM Group and Control GroupCharacteristicMPM Group(n¼40)Control Group(n¼448) P_Age, y 62.58_11.32 55.69_11.73 <0.001< span="">Sex 0.275Male 36 369Female 4 79Family historyof malignancies0.516No 31 371Yes 9 77HBsAg status 0.013Negative 17 106Positive 23 342HCV antibody 0.072Negative 33 412Positive 7 36ALT, U/L 0.406_40 20 258>40 20 190GGT, U/Ly 0.317_67 26 235>67 14 187AFP, ng/mLy 0.001_20 26 161>20 14 260Tumor size, cm 0.243_5 19 260>5 21 188Tumor location 0.729Right liver 29 297Left liver 7 85Both 4 66Multiple tumors 0.830No 34 366Yes 6 82Vascular invasion 0.820No 35 384Yes 5 64Edmondson grade 0.606I–II 24 290III–IV 16 158Cirrhosis 0.283No 15 131Yes 25 317Nodal status 0.403Negative 39 443Positive 1 5TNM staging 0.124I–II 36 428III–IV 4 20‘‘Bold’’ value means the ‘‘P’’ value is less than 0.05.AFP¼alfa-fetoprotein, ALT¼alanine transaminase, GGT¼gammaglutamyltransferase, HBsAg¼hepatitis B virus surface antigen,HCV¼hepatitis C virus, MPM¼multiple primary malignancies._Fisher’s exact test or Pearson’s x2-test.y1.34% and 6.25% data in control group was missing.Xu et al Medicine _ Volume 95, Number 17, April 20164 | www.md-journal.comCopyright # 2016 Wolters Kluwer Health, Inc. All rights reserved.were HCVt and did not significantly differ from the HCC-onlygroup. We are not surprised at the difference in HBsAg infectionbetween the 2 groups, as reasons for HCC development inMPM patents may be more complex than those for the HCConlygroup, although HBV and HCV infections were regardedas major causes for HCC.MPM has been attributed to iatrogenic, environmental, andhereditary factors.17 Iatrogenic factors, such as anticancer treatmentsor radiation therapy, were considered as causes of MPMtumors. Reportedly, about 40% of patients with metachronousMPM had histories of receiving anticancer treatments or radiationtherapy to attempt to cure their first cancers and consequentlydeveloped secondary tumors following their initialtreatment.18 In our series, 37.9% (11/29) of MPM patients in themetachronous group had received chemotherapy or radiotherapy.Although we might have further considered the effects ofradiation or chemical regimens, age at radiation exposure, andsubsequent treatments, no clear differences were observedbecause of insufficient information.Hereditary factors may be another cause of MPM tumors.Family history of malignancies, which is regarded as a riskfactor for HCC, may also portend HCC development as asecond malignancy. In our MPM group, 22.5% (9/40) hadimmediate family members with histories of cancer, whichwas similar to the patients in the HCC-only group (17.2%77/371). We hypothesize that hereditary factors play a rolein the process, but not solely in MPMs.Aging is an important etiological factor in MPM patients.Using the Osaka Cancer Registry data, Tabuchi et al19 reportedthat 10-year cumulative risk of metachronous second primarycancer in Japanese male patients was 10.2% at 50 to 59 years ofage, 16.2% at 60 to 69 years of age, and 21.8% at 70 to 79 yearsof age. In the present study, the mean age of HCC diagnosisin MPM patients was 62.58_11.32 years, which was significantlyolder than that of the HCC-only control (55.69_11.73years). Furthermore, the mean ages of diagnosis did not significantlydiffer between the synchronous and metachronousgroups (60.18_8.86 vs. 63.48_12.14 years, P¼0.418). Thisimplies that older people have higher risks of developingsecond malignancies, without the choice for synchronous ormetachronous.Other risk factors such as BMI, immune status, andbehavior change after the first primary malignancy may alsocontribute to HCC development in MPM patients,20 but moredetailed investigation is needed. In most cases, inherited,iatrogenic, or viral factors are implicated; in other cases a clearetiopathogenesis is difficult to find, especially for synchronousMPMs. In our study, 2 synchronous HCC lesions withoutcirrhosis in background liver were surprisingly diagnosedby pathology after surgery for what were thought to be livermetastasis. One extra-hepatic synchronous tumor was unexpectedlyfound during the surgery, which was regardedas a benign lesion. Thus, the mechanism still needs furtherclarification.TABLE 4. Univariate Analysis of Survival Risk Factors for MPMPatientsCharacteristic3-YearSurvivalrate (%)5-YearSurvivalrate (%) PAge, y_ 0.106_62.5 (n¼18) 50.0 25.0>62.5 (n¼22) 72.7 48.0Gender 0.550Male (n¼36) 61.0 35.9Female (n¼4) 75.0 50.0HBsAg status 0.404Negative (n¼17) 70.1 44.6Positive (n¼23) 56.5 31.4HCV-Ab 0.448Negative (n¼33) 63.5 39.4Positive (n¼7) 57.1 28.6ALT, U/L 0.699_40 (n¼20) 64.6 26.4>40 (n¼20) 60.0 45.0GGT, U/L 0.009_67 (n¼26) 72.9 47.5>67 (n¼14) 42.9 17.1AFP, ng/mL 0.023_20 (n¼26) 76.7 43.3>20 (n¼14) 35.7 26.8Cirrhosis 0.280No (n¼15) 80.0 43.6Yes (n¼25) 56.0 34.5Tumor size, cm < 0.001_5 (n¼19) 89.5 67.7>5 (n¼21) 37.5 6.3Multiple tumors 0.727No (n¼34) 61.6 38.1Yes (n¼6) 66.7 33.3Vascular invasion 0.021No (n¼35) 62.7 43.2Yes (n¼5) 60.0 0Edmondson grade 0.839I–II (n¼24) 58.8 39.2III–IV (n¼16) 68.8 34.4Synchronous or metachronous 0.044Synchronous (n¼11) 45.5 13.6Metachronous (n¼29) 68.8 45.9Treatments for extra-hepaticprimary malignancies0.614Curative therapy (n¼26) 67.6 49.4Palliative therapy (n¼4) 50.0 50.0‘‘Bold’’ value means the ‘‘P’’ value is less than 0.05.AFP¼alfa-fetoprotein, ALT¼alanine transaminase, GGT¼gammaglutamyltransferase, HBsAg¼hepatitis B virus surface antigen,HCV¼hepatitis C virus, MPM¼multiple primary malignancies._Patients’ age was divided by the median age.TABLE 5. Cox Analysis of Survival Risk Factors for MPMPatientsCharacteristic P HR (95% CI)GGT level 0.501 1.459 (0.486–4.377)AFP level 0.459 1.409 (0.527–3.772)Tumor size <0.001 1.455 (1.184–1.788)Vascular invasion 0.065 3.504 (0.927–13.243)Synchronous or Metachronous 0.111 0.438 (0.159–1.207)‘‘Bold’’ value means the ‘‘P’’ value is less than 0.05.AFP ? alfa-fetoprotein, GGT ? gamma glutamyltransferase.Medicine _ Volume 95, Number 17, April 2016 Multiple Primary Malignancies Associated With Hepatocellular CarcinomaCopyright # 2016 Wolters Kluwer Health, Inc. All rights reserved. www.md-journal.com | 5Information about common sites of extra-hepatic malignanciesmay improve early detection in high-risk individuals.21Gastric cancer has been reported as the most common extrahepaticmalignancy among MPM patients with HCC byTakayasu et al,22 along with colorectal cancer by Ferna´ndez-Ruiz et al,23 and nasopharynx cancer by Zeng et al.24 Unlikethese previous findings, our study showed that the most commonextra-hepatic malignancy was lung, followed by colorectaland thyroid. This circumstance may be partly attributable todifferent regions from which the study subjects were selected,for the most common forms of extra-hepatic malignancies weresimilar to the most common tumor types in China;25–28 andpartly to the wide variation of multiple cancer distribution,which may occur as a result of random chance. Contrary to ourexpectation, screening for other possible malignancies in cancersurvivors based on the most common sites is difficult because ofthe variable distribution of the extra-hepatic malignancy andany enrichment patterns can hardly have been proven bystatistics yet. Establishment of a pair-wise association withHCC requires a more systematic and controlled approach.Previous studies indicate that patients who initially presentedwith thyroid, urinary bladder, prostate, cervical, anduterine cancers were more liable to develop second malignancies,whereas those with hepatic cancers rarely developed asecond malignancy. They hypothesized that this was, as HCChas a poor prognosis, HCC patients did not survive long enoughto develop second primaries.29 HCC was among the four cancersites with the lowest survival rates and consequently, the shortestduration of follow-up.30 However, about 40% (11/29) of ourmetachronously diagnosed patients were in the post group. Thepoor prognosis of HCC patients apparently does not affect theincidence of another primary tumor occurrence, and the possibilityof developing extra-hepatic malignancies in HCC patientsshould not be ignored. Only the obviously longer interval timeof the prior group can be explained partly by poor OS for HCC.Our MPM patients with interval times longer than 72 monthswere all in the prior group (Figure 1). In view of this pattern,physicians must consider the onset of HCC for each neoplasm,even many years after first diagnosis.No consensus currently exists for a method of calculatingthe survival rate of MPM patients. Earlier researchers recommendedbasing the rate from the diagnosis of the final malignancytumor, while others suggest calculating survival from thediagnosis of the first tumor, to account for the increased risk ofmalignancy during the first survival period.31 We focused onsurvival time after surgeries for HCC because most MPMpatients died of HCC-related causes, which may indicate thatMPM prognosis is largely determined by survival time after theHCC surgery. However, this may avoid the bias brought bylonger intervals between MPM diagnoses, which could indicatea longer survival time.Survival of MPM patients is reportedly similar to that ofpatients with single primary tumors.23,32,33 We had the samefindings for post-surgical survival time (Figure 2C). Further, wefound no significant difference in surgery-related parameters,such as surgery time and amount of bleeding, between MPMand control group. We speculate that a history of extra-hepatictumor is not a direct obstacle to HCC resection. MPMitself doesnot necessarily indicate a poor prognosis, as long as adequatediagnosis and management are performed. However, HCCrelatedcauses predominantly lead to MPM patients’ deaths;only 38.6% of patients in this study were still alive 5 years aftertheir liver surgeries.Male sex and old age have been shown by several studies tobe risk factors for shorter survival in MPM.34 However, wefound no statistical difference for OS in these terms. In thepresent study, serum GGT level, AFP level, tumor size, vascularinvasion, and synchronous or metachronous diagnosis led todistinct outcomes. We verified these results with a Cox multivariatemodel, which only found tumor size, as a pathologicalfeature, to be a significant independent risk factor for survival.This is an important new observation for patients who survivedtheir first primary malignancy. Early detection and surgery forHCC would help improve OS in these patients.Although the metachronous and synchronous groups significantlydiffered in OS (Figure 2A), metachronous or synchronousdiagnoses were not independent OS factors inmultivariate analysis. Metachronous malignant lesions werediscovered because of careful follow-up of the first malignancy,during which extensive surveillance is carried out to locatepossible metastases. This may explain why HCC lesions inmetachronous group (mainly in the prior group) were found assmaller tumors than in the synchronous group, which may offerlonger survival. Moreover, no significant difference in OS wasfound between the prior and post groups, which demonstratethat whether extra-hepatic malignancy was the initial or secondarymalignancy did not influence OS after surgeries forHCC (Figure 2B). Another hypothesis is that as more timeelapses between the 2 primary malignancies, the better theprognosis. Although the post group has a longer median periodthan the synchronous group before diagnosis of second malignancies,their OS rates did not significantly differ (P¼0.239).We found no relationship between second primary tumordevelopment and MPM survival rate of MPM patients.FIGURE 2. Comparisons of Kaplan–Meier curves between synchronous and metachronous groups (A); prior and post groups (B), andMPM and control groups (C). MPM ? multiple primary malignancies.6 | www.md-journal.comCopyright # 2016 Wolters Kluwer Health, Inc. All rights reserved.In summary, MPMs associated with HCC is rare. Our studyprovides the largest sample size of MPM patients ever, receivingradical resections for HCC. MPM patients were more likelyto die of HCC-related causes even after receiving radicalresection for HCC. Tumor size, rather than MPM itself, wasthe only independent predictive factor for OS in MPM patients.Follow-up for patients recovering from a first malignancy mustbe strictly observed, which could improve their chances forlong-term survival. Because of the complex etiology and thevariety ofMPMcancer distributions, HCC should be consideredas a potential second primary for every cancer survivors, even ifnot infected by HBV. Additionally, HCC patients, especiallyelderly ones, all malignancies must be considered risks ofsecond tumor.This study is subject to the limitations inherent in retrospectivework with observation data collected at the specificpoint. It also represents the experience of a single tertiary referralcenter, and might not be generalized. The etiology of MPMsremains unclear, because risk factors known to be important toetiology, such as the details of chemotherapy or radiation therapy,could hardly be estimated in this study. Limitations of our studyalso include the confined sample size, although we have thelargest sample size.Alarger,multi-center study of patients fromamulti-geographic patient base would be more conclusive.