ORIGINAL ARTICLE Tumor-Strom Rtio Is n Independent Predictor for Survivl in Esophgel Squmous Cell Crcinom Ki Wng, MD,* Wei M, MD,* Jinbo Wng, MD,* Ling Yu, MD, Xiomei Zhng, MD, Zhenbo Wng, MD, Bingxu Tn, MD,* Nn Wng, MD,* Bing Bi, MD,* Shengsi Yng, MD,* Houqing Liu, MD, Shengjie Zhu, MD, nd Yufeng Cheng, MD, PhD* Objective: Tumor-strom rtio (TSR) hs been identified s new nd prcticble prognostic histologicl chrcteristic of solid tumors. The im of this study ws to evlute the prognostic vlue of TSR in resected esophgel squmous cell crcinom (ESCC). Methods: A totl of 95 ptients who underwent esophgectomy for ESCC were included in this study. TSR ws ssessed visully on the hemtoxylin-eosin stined tissue sections of surgicl specimens by two independent observers. Ptients with more thn 50% intrtumor strom were quntified s the strom-rich group nd those with less thn 50% s the strom-poor group. Results: No significnt differences were observed in ptient, tumor, nd tretment chrcteristics between the strom-rich nd strom-poor groups. The 3-yer overll survivl nd disese-free survivl rtes were 64% nd 57%, respectively, in the strom-poor group, nd 23% nd 23%, respectively, in the strom-rich group. Both 3-yer overll nd disese-free survivl rtes in the strompoor group were significntly better thn those in the strom-rich group (p < 0.01). In multivrite nlysis, TSR ws identified s highly significnt prognostic fctor for 3-yer overll survivl (hzrd rtio 3.450; p = 0.001) nd 3-yer disese-free survivl (hzrd rtio 2.995; p = 0.001), independent of ptnm stge nd rdiclity of the primry tumor. Conclusion: Strom-rich tumors were ssocited with poor prognosis nd n incresed risk of relpse, which my serve s new prognostic histologicl chrcteristic in ESCC. TSR is simple nd quick to determine, is reproducible, nd could be esily incorported in routine histologicl evlution. Key Words: Squmous cell crcinom, Esophgel neoplsms, Survivl, Tumor-strom rtio (J Thorc Oncol. 2012;7: 1457 1461) *Oncology Center, Qilu Hospitl of Shndong University, Jinn, People s Republic of Chin; Affilited Hospitl of Binzhou Medicl College, Ynti, People s Republic of Chin; Affilited Hospitl of Binzhou Medicl College, Binzhou, People s Republic of Chin; nd Deprtment of Oncology, Wendeng Centrl Hospitl of Weihi, Weihi People s Republic of Chin. Disclosure: The uthors declre no conflicts of interest. Address for Correspondence: Yufeng Cheng, MD, PhD, Oncology Center, Qilu Hospitl of Shndong University, 107 West Wenhu Rd, Jinn, 250012, P.R. Chin. E-mil: qilucyf@126.com. Copyright 2012 by the Interntionl Assocition for the Study of Lung Cncer ISSN: 1556-0864/12/0709-1457 Esophgel crcinom is the eighth most common cncer nd the sixth leding cuse of cncer deth in the world. 1 There re two mjor histologicl types of esophgel crcinom: esophgel squmous cell crcinom (ESCC) nd denocrcinom with considerbly vried epidemiologicl fetures. In Chin nd other Est Asin countries, more thn 90% of cses re ESCC, wheres denocrcinom is more common in the United Sttes nd Europen countries. Tumors re complex tissue composed of crcinom cells nd surrounding strom. The prognostic fctors of ESCC include tumor loction, depth of tumor invsion, number of involved lymph nodes, histologicl grde, nd ptnm stge. To dte, it is incresingly cknowledged tht tumor strom is indispensble for cncer initition, progression, nd metstsis, nd the stroml elements of tumor hold prognostic potentil. Recently, tumor-strom rtio (TSR) ws proved to be new prognostic fctor in esophgel denocrcinom, erly brest crcinom, nd colon crcinom. 2 4 To our knowledge, the prognostic vlue of TSR remined to be explored for ESCC. The objective of this study ws to evlute the prognostic vlue of TSR in ESCC nd its reltionship with other prognostic fctors. PATIENTS AND METHODS Ptients nd Tissue Smples From the dtbse of the Qilu Hospitl of Shndong University, we selected ll ptients with ESCC who underwent resection with curtive intent t the Deprtment of Thorcic Surgery of the Qilu Hospitl from 1 Jnury to 31 December 2007. Ptients treted with neodjuvnt therpy, which could interfere with the evlution of TSR were excluded, s were ptients who died within 30 dys fter surgery. Ptient, tumor, nd tretment chrcteristics were retrieved from the Medicl Records Room. Both originl pthologicl reports nd hemtoxylin nd eosin (HE) stined sections from the primry tumor were retrieved from the Deprtment of Pthology. All ptients signed informed consent to this study, nd the protocol ws pproved by the Ethics Committee of Qilu Hospitl of Shndong University. Histopthologicl Protocol The 5-µm HE-stined sections from the primry tumor were routinely nlyzed on microscopic exmintion. In cse Journl of Thorcic Oncology Volume 7, Number 9, September 2012 1457
Wng et l. Journl of Thorcic Oncology Volume 7, Number 9, September 2012 of tumor heterogeneity, strom-rich res were considered to be of worse prognostic vlue nd therefore deemed decisive. 5 In generl, res rich in strom were found ner the site of deepest microscopic infiltrtion. The surrounding stroml tissue not contining tumor cells ws considered not to be connected with the tumor. Using 4 microscope objective (40 totl mgnifiction), we selected the most invsive tumor re for further evlution. Subsequently, microscopicl fields, where both strom nd tumor were present, nd tumor cells were visulized on ll sides nd were scored with 10 objective (100 totl mgnifiction). The ssessment ws done on the bsis of the nlysis of t lest one microscopic field. The estimte ws then recorded s the TSR. With this protocol, tumor sections were independently ssessed by two investigtors. A third pthologist s decision ws decisive when the two observers disgreed. A 50% cutoff vlue ws tken s previously determined in colon cncer, which ws lso confirmed in esophgel denocrcinom. 3 TSR ws defined s strom poor (the proportion of strom <50%) or strom rich (the proportion of strom 50%). Tumor node metstsis stging ws done ccording to the Americn Joint Committee on Cncer Cncer Stging Mnul (7th ed, 2010). Follow Up Follow-up dt were collected until deth or Mrch 2011. All ptients hd regulr follow-up schedule including complete history nd physicl exmintion every 3 months during the first 2 yers fter surgery nd every 6 months therefter. Routine rdiologicl exmintions were performed when necessry. Sttisticl Anlysis Differences in ptient, tumor, nd tretment chrcteristics were ssessed using the Mnn Whitney test for continuous vribles nd the χ 2 test for ctegoricl vribles. The Cohen s κ coefficient ws used to nlyze the vribility between the two investigtors. The Kpln Meier method nd log-rnk test were used for nlysis nd comprison of survivl curves. For the nlysis of 3-yer overll survivl, events were defined s deth from ny cuse. For the nlysis of 3-yer disese-free survivl, events were defined s first loco-regionl or distnt tumor relpse or deth from ny cuse. The Cox proportionl hzrds model ws used to determine the hzrd rtio (HR) of vribles on 3-yer overll survivl nd 3-yer disese-free survivl in univrite nd multivrite nlysis. The results re given s HRs with their 95% confidence intervl (CI). p vlues less thn 0.05 were considered sttisticlly significnt. Dt were nlyzed using sttisticl pckge SPSS version 17.0 (SPSS Inc., Chicgo, IL). RESULTS Clinicopthologicl Fetures Ninety-five ptients (82 men nd 13 women) were included in this study. The medin ge of the ptients ws 60 (rnge, 42 77) yers t the dte of surgery. The medin follow-up time ws 40 (4 50) months. Clinicopthologicl nd tretment chrcteristics of ptients re shown in Tble 1. TABLE 1. Ptient, Tumor nd Tretment Chrcteristics (Grouped by Tumor Strom Rtio) Chrcteristics Totl Strom-Poor Strom-Rich (N = 95) (n = 65) (n = 30) No. % No. % No. % Sex Men 82 86.3 58 89.2 24 80.0 Women 13 13.7 7 10.8 6 20.0 0.224 Medin ge (rnge) 60.0 (42 77) 60.0 (42 77) 60.5 (45 77) 0.128 Medin tumor length 38 (9 130) 40 (9 85) 38 (9 130) 0.926 in mm (rnge) Tumor loction Upper 10 10.5 7 10.8 3 10.0 Middle 56 59.0 41 63.0 15 50.0 0.387 Low 29 30.5 17 26.2 12 40.0 pt sttus pt1 10 10.5 9 13.8 1 3.3 pt2 18 19.0 15 23.1 3 10.0 0.060 pt3/4 67 70.5 41 63.1 26 86.7 pn sttus pn0 56 58.9 38 58.5 18 60.0 pn1 25 26.3 19 27.7 8 23.3 pn2 11 11.6 7 10.8 4 13.3 0.964 pn3 3 3.2 2 3.0 1 3.4 Lymph node rtio <0.2 73 76.8 53 81.5 20 66.7 0.2 22 23.2 12 18.5 10 33.3 0.110 ptnm stge I 13 13.7 11 16.9 2 6.7 II 48 50.5 32 49.2 16 53.3 0.394 III 34 35.8 22 33.8 12 40.0 Differentition grde Well 28 29.5 20 30.8 8 26.7 Moderte 40 42.1 27 41.5 13 43.3 0.918 Poor 27 28.4 18 27.7 9 30.0 Rdiclity R0 92 96.8 63 96.9 29 96.7 R1 2 2.1 2 3.1 0 0 0.212 R2 1 1.1 0 0 1 3.3 Adjuvnt therpy No 47 49.5 35 53.8 12 40.0 Yes 48 50.5 30 46.2 18 60.0 0.210 pt, pthologicl tumor stge; pn, pthologicl node stge; ptnm, tumor node metstsis. p < 0.05 ws considered significnt. Tumor-Strom Rtio in ESCC With 4 nd 10 objectives, routine HE stined sections from the primry tumors were nlyzed for the presence of stroml involvement. TSR ws ssessed on one section derived from the most invsive prt of the tumor (Fig. 1). Estimtion of the TSR ws performed successfully in ll tumors. Assessed by two independent reserchers, 65 tumors were strom poor nd 30 were strom rich. Cohen s κ p 1458 Copyright 2012 by the Interntionl Assocition for the Study of Lung Cncer
Journl of Thorcic Oncology Volume 7, Number 9, September 2012 TSR is n Independent Predictor in ESCC FIGURE 1. Hemtoxylin nd eosin stined 5 µm sections of eso phgel squmous cell crcinom (originl mgnifiction 100). A, Exmple of strom-rich (strom rtio 50%). B, Exmple of strom-poor (strom rtio < 50%). coefficient reveled n lmost perfect greement between two reserchers (κ = 0.84). Correltion of TSR With Other Prognostic Fctors Tble 1 shows ptient, tumor, nd tretment chrcteristics for the strom-rich nd the strom-poor groups. There were no significnt differences between the two groups. Follow up ws complete. The 3-yer overll survivl rte nd disese-free survivl rte were 64% nd 57%, respectively, in the strompoor group nd 23% nd 23%, respectively, in the strom-rich group. Medin overll survivl for ptients in the strom-poor group ws 39 months (95% CI 36 43 months) compred with 24 months (95% CI 18 29 months) for ptients in the stromrich group. Survivl curves re shown in Fig. 2. The difference of survivl curves between the two groups remined sttisticlly significnt. In the Cox univrite nd multivrite nlysis of 3-yer survivl, the HR of TSR ws 3.548 (95% CI 1.977 6.365; p = 0.001) nd 3.450 (95% CI 1.913 6.222; p = 0.001), respectively. In the Cox univrite model, TSR, lymph node sttus, lymph node rtio, ptnm stge, depth of tumor invsion (T3/4), nd rdiclity of resection were FIGURE 2. Kpln-Meier curves for survivl of ll 95 ptients with esophgel squmous cell crcinom (strom poor versus strom rich). Both 3-yer overll survivl (A) nd disesefree survivl (B) between two groups remined sttisticlly significnt. significntly relted to 3-yer overll survivl. In the Cox univrite nd multivrite nlysis of disese-free survivl, the HRs for TSR were 2.635 (95% CI 1.526 4.548; p = 0.001) nd 2.995 (95% CI 1.706 5.260; p = 0.001), respectively. TSR ws lso n independent prognostic vrible for 3-yer disese-free survivl nd node sttus nd rdiclity of resection (Tbles 2 nd 3). DISCUSSION The present investigtion shows tht TSR is new independent prognostic fctor for ESCC. Strom-rich tumors were ssocited with poor prognosis nd n incresed risk of relpse. The use of TSR s prognostic fctor hs been introduced by previous studies.2 4 After nlysis of 122 ptients with stge I to III colon crcinom, Mesker et l.4 found tht ptients with TSR less thn 50% showed significnt worse overll nd disese-free survivls, which suggested tht Copyright 2012 by the Interntionl Assocition for the Study of Lung Cncer 1459
Wng et l. Journl of Thorcic Oncology Volume 7, Number 9, September 2012 TABLE 2. Cox Univrite Anlysis for 3-Yer Survivl in 95 Ptients of Esophgel Squmous Cell Crcinom 3-Yr Overll Survivl Univrite Anlysis 3-Yr Disese-Free Survivl HR 95% CI p b HR 95% CI p b TSR 0.001 0.001 Strom-poor 1.000 Ref. 1.000 Ref. Strom-rich 3.548 1.977 6.365 0.001 2.635 1.526 4.548 0.001 Differentition grde 0.565 0.469 Well 1.000 Ref. 1.000 Ref. Moderte 1.444 0.696 2.996 0.324 1.263 0.640 2.496 0.501 Poor 1.450 0.658 3.195 0.356 1.573 0.763 3.243 0.220 Tumor loction 0.362 0.365 Upper 1.000 Ref. 1.000 Ref. Middle 0.632 0.261 1.529 0.309 0.780 0.326 1.865 0.577 Low 0.490 0.184 1.307 0.490 0.537 0.205 1.408 0.206 pt sttus 0.054 0.362 pt1 1.000 Ref. 1.000 Ref. pt2 4.254 0.523 34.581 0.176 1.104 0.361 3.377 0.862 pt3/4 7.741 1.061 56.452 0.044 1.676 0.658 4.269 0.279 pn sttus 0.044 0.145 pn0 1.000 Ref. 1.000 Ref. pn1 1.771 0.899 3.488 0.098 1.316 0.692 2.501 0.403 pn2 2.522 1.115 5.702 0.026 1.771 0.804 3.901 0.156 PN3 3.501 1.043 11.752 0.043 3.439 1.038 11.389 0.043 Lymph node rtio 0.025 0.183 <0.2 1.000 Ref. 1.000 Ref. 0.2 2.029 1.094 3.761 0.025 1.501 0.825 2.731 0.183 ptnm stge 0.015 0.118 I 1.000 Ref. 1.000 Ref. II 7.981 1.075 59.258 0.042 1.471 0.599 3.607 0.400 III 13.404 1.807 99.436 0.011 2.313 0.934 5.728 0.070 Rdiclity 0.048 0.068 R0 1.000 Ref. 1.000 Ref. R1 3.424 0.813 14.432 0.094 2.392 0.575 9.955 0.231 R2 7.154 0.933 54.842 0.058 8.350 1.076 64.766 0.042 Adjuvnt therpy 0.454 0.254 No 1.000 Ref. 1.000 Ref. Yes 1.248 0.699 2.231 0.454 1.374 0.796 2.372 0.254 pt, pthologicl tumor stge; pn, pthologicl node stge; ptnm: tumor node metstsis; TSR, tumor-strom rtio; HR, hzrd rtio; CI, confidence intervl; Ref., reference. Anlysis ws performed using the Cox proportion hzrd model. b p <0.05 ws considered significnt (indicted in bold). TSR could serve s n independent prmeter for confident prediction of clinicl outcome in erly-stge colon cncer. Recently, TSR ws lso confirmed to be new nd prcticble prognostic tumor chrcteristic in esophgel denocrcinom nd erly brest cncer. 2 4 Our hypothesis ws tht TSR might lso be n importnt prognostic prmeter for ESCC. The TABLE 3. Multivrite Cox Anlysis for 3-Yr Survivl in 95 Ptients of Esophgel Squmous Cell Crcinom pt sttus (T3/4) pn sttus (N3) TSR (Stromrich) Lymph node rtio 0.2 ptnm stge(iii) Rdiclity (R1-2) 3-Yr Overll Survivl Multivrite Anlysis 3-Yr Disese-Free Survivl HR 95% CI p b HR 95% CI p b 2.727 0.330 22.553 0.352 0.988 0.319 3.059 0.983 2.715 0.693 10.630 0.152 4.652 1.393 15.541 0.012 3.450 1.913 6.222 0.001 2.995 1.706 5.260 0.001 1.280 0.542 3.024 0.573 0.794 0.333 1.895 0.603 11.560 1.556 85.873 0.017 1.990 0.596 6.647 0.264 4.738 1.395 16.090 0.013 4.125 1.244 13.679 0.020 pt, pthologicl tumor stge; pn, pthologicl node stge; ptnm, tumor node metstsis; TSR, tumor-strom rtio; HR, hzrd rtio; CI, confidence intervl. Anlysis ws performed using the Cox proportion hzrd model. b p < 0.05 ws considered significnt. optiml threshold level of TSR ws determined on the bsis of mximum discriminting power for overll survivl nd disese-free survivl. The 50% cutoff vlue ws the most representtive. In the study, we found tht 3-yer survivl rte nd disese-free survivl rte were 64% nd 57%, respectively, in the strom-poor group, nd 23% nd 23%, respectively, in the strom-rich group. TSR ws of prognostic vlue by both univrite nd multiverte nlysis. Mny prognostic fctors hve been found for ESCC, such s the number nd the rtio positive lymph nodes, 6 ptnm stge, tumor loction, depth of tumor invsion, nd histologicl grde. All these fctors hve been included in our nlysis. In the study, we found tht lymph node sttus, positive lymph node rtio, ptnm stge, depth of tumor invsion (T3/4), nd rdiclity of resection were significntly relted to 3-yer overll survivl in univrite nlysis, nd ptnm stge nd rdiclity of resection were lso independent prognostic fctors in multivrite nlysis. For 3-yer disese-free survivl, rdiclity of resection nd pn3 sttus (metstsis in seven or more regionl lymph nodes) were independent prognostic fctors through multivrite nlysis. Our results were similr to those of previous studies. 7 Tumor tissue is composed of both crcinom cells nd stroml cells recruited from norml tissue. In norml tissue, the strom my ctully ct s brrier in tumorigenesis by constrining tumor cell prolifertion. In tumor tissue, however, stroml components the min prt of tumor microenvironment could fcilitte the process of tumor progression. 8 The mechnism underlying tumor-promoting effect of strom is still not fully understood, which my be ssocited with n incresed number of fibroblsts (higher strom rtio), reconstruction of the extrcellulr mtrix, incresed microvessel nd microlymphtic density, recruitment of inflmmtory cells etc. Recently, the so-clled cncer-ssocited fibroblsts (CAFs) composed of the mjor 1460 Copyright 2012 by the Interntionl Assocition for the Study of Lung Cncer
Journl of Thorcic Oncology Volume 7, Number 9, September 2012 TSR is n Independent Predictor in ESCC cellulr components of tumor strom were found to hve predominnt role in tumor growth nd progression. Quiescent fibroblsts ctivted in tumor strom re key regultors of the prcrine signling between strom nd cncer cells. 9 CAFs, consisting of both fibroblsts nd myofibroblsts, re frequently observed in the strom of humn crcinom nd secrete vriety of soluble fctors such s trnsforming growth fctor bet 1 (TGF-β1), stroml cell-derived fctor 1 nd other soluble fctors secreted by CAFs, tht ct in prcrine mnner nd ffect not only cncer cells but lso other cell types present in the strom. 10 CAFs could produce vrious growth fctors, cytokines, nd extrcellulr mtrix proteins to promote ngiogenesis, which is essentil for tumor growth nd progression. 11 Tumor immunity is lso medited vi cytokines secreted by CAFs. Blsmo et l. 12 found tht CAFs could modulte immune cell type nd ntitumor ctivity. The CAFs ply n importnt role in some spects of cncer biology including trnsformtion, progression, tumor growth, nd drug resistnce, 13 nd their presence in lrge numbers is often ssocited with the development of highgrde mlignncies nd poor prognosis. 14 Our results indicted tht incresed mount of stroml proportion in ESCC cn be connected with poor prognosis independent of other prognostic fctors. In ddition, TSR cn be esily ssessed nd performed during routine pthologicl exmintion. 15 Although our study popultion ws homogenous, the study hs its shortcomings. This ws retrospective study with reltively smll smple size nd the follow-up time is short. It is of greter vlue to conduct prospective biopsy study with lrger smple size in the future. In conclusion, our findings indicte tht TSR is new prognostic fctor for ESCC. TSR cn be determined esily during routine pthologicl exmintion on HE-stined sections of the resected tumor. Crcinom hs long been viewed s disese of trnsformed epithelil cells, nd tretment strtegies focused only on tumor cells. The study of tumor strom hs opened gret opportunities for future cncer therpy. Trgeting components of the tumor microenvironment hve gret potentil in the clinicl prctice, prticulrly when used in combintion with other therpeutic gents. ACKNOWLEDGMENTS The uthors thnk Dr. Zhng TG (pthologist in Deprtment of Pthology, Shndong University School of Medicine, Chin) for his expert suggestions nd technicl ssistnce. REFERENCES 1. Jeml A, Bry F, Center MM, Ferly J, Wrd E, Formn D. Globl cncer sttistics. CA Cncer J Clin 2011;61:69 90. 2. de Kruijf EM, vn Nes JG, vn de Velde CJ, et l. Tumor-strom rtio in the primry tumor is prognostic fctor in erly brest cncer ptients, especilly in triple-negtive crcinom ptients. Brest Cncer Res Tret 2011;125:687 696. 3. Courrech Stl EF, Wouters MW, vn Sndick JW, et l. The stroml prt of denocrcinoms of the oesophgus: does it concel trgets for therpy? Eur J Cncer 2010;46:720 728. 4. Mesker WE, Liefers GJ, Junggeburt JM, et l. Presence of high mount of strom nd downregultion of SMAD4 predict for worse survivl for stge I-II colon cncer ptients. Cell Oncol 2009;31:169 178. 5. de Kruijf EM, vn Nes JG, vn de Velde CJ, et l. Tumor-strom rtio in the primry tumor is prognostic fctor in erly brest cncer ptients, especilly in triple-negtive crcinom ptients. Brest Cncer Res Tret 2011;125:687 696. 6. Liu YP, M L, Wng SJ, et l. Prognostic vlue of lymph node metstses nd lymph node rtio in esophgel squmous cell crcinom. Eur J Surg Oncol 2010;36:155 159. 7. Smit JK, Pultrum BB, vn Dullemen HM, Vn Dm GM, Groen H, Plukker JT. Prognostic fctors nd ptterns of recurrence in esophgel cncer ssert rguments for extended two-field trnsthorcic esophgectomy. Am J Surg 2010;200:446 453. 8. Bissell MJ Rdisky D. Putting tumours in context. Nt Rev Cncer 2001;1:46 54. 9. Lorusso G Rüegg C. The tumor microenvironment nd its contribution to tumor evolution towrd metstsis. Histochem Cell Biol 2008;130:1091 1103. 10. Mssgué J. TGFbet in Cncer. Cell 2008;134:215 230. 11. Kojim Y, Acr A, Eton EN, et l. Autocrine TGF-bet nd stroml cell-derived fctor-1 (SDF-1) signling drives the evolution of tumorpromoting mmmry stroml myofibroblsts. Proc Ntl Acd Sci USA 2010;107:20009 20014. 12. Blsmo M, Scordmgli F, Pietr G, et l. Melnom-ssocited fibroblsts modulte NK cell phenotype nd ntitumor cytotoxicity. Proc Ntl Acd Sci USA 2009;106:20847 20852. 13. Udgw T Wood M. Tumor-stroml cell interctions nd opportunities for therpeutic intervention. Curr Opin Phrmcol 2010;10:369 374. 14. Shimod M, Mellody KT, Orimo A. Crcinom-ssocited fibroblsts re rte-limiting determinnt for tumour progression. Semin Cell Dev Biol 2010;21:19 25. 15. Courrech Stl EF, Smit VT, vn Velthuysen ML, et l. Reproducibility nd vlidtion of tumour strom rtio scoring on oesophgel denocrcinom biopsies. Eur J Cncer 2011;47:375 382. Copyright 2012 by the Interntionl Assocition for the Study of Lung Cncer 1461