In!. J. Radaron Oncolqq Bwl Phw Vol. 19. pp. 1143-l 149 Printed in the U.S.A. All n&s reserved.
Copyright
0360.3016/90 $3.00 + .I0 0 I990 Pergamon Press plc
??Original Contribution
PROGNOSTIC
SIMON
FACTORS OF NASOPHARYNGEAL A MULTIVARIATE ANALYSIS
Guo JENG TANG, MING WAI
SHEN MAN
M.D.,
CHEN,
M.D.,*
LEUNG,
M.D.*
Chang-Gung
M.S.,*
FANG
CHUANG
JEN LIN, CHI
AND JI HONG
CARCINOMA:
M.D.,
LIAW,
M.D.,+
HONG,
M.D.*
PH.D.,*
Memorial Hospital, Taipei, Taiwan
Between 1979 and 1985,561 patients with nasopharyngeal carcinoma were reviewed to determine prognostic factors that may influence survival. Sex (p = 0.294) and histopathology (p = 0.677) had no correlation to the acturial survival, whereas the site of cervical metastasis (p = 0.001) and the radiation doses to the nasopharynx and regional lymph nodes (p = 0.03) were both significant when one used univariate analyses. Cox’s multivariate regression model revealed that the presence rather than the site of distant metastases was the single most important independent factor influencing the treatment outcome (p < 0.0001). The addition of chemotherapy, on the other hand, did not show a survival benefit even when one took available confounding factors into account. There are, however, survival advantages associated with: (a) young age (140 years), (b) asymptomatic status, (c) Stage I or II lesions, and (d) biopsy via nasopharynx instead of neck nodes. These favorable prognostic factors may be used for therapeutic guidance and end-result reporting. Nasopharyngeal
carcinoma, Radiotherapy,
Prognostic factors.
INTRODUCI’ION
METHODS
Nasopharyngeal carcinoma (NPC) occurs with a relatively high frequency among head and neck cancers in southeast Asia, where it has been implicated with Epstein-Barr (EB) viral infection ( 10, 24) due to an elevated serum level of E-B antibody (8). The role of radiotherapy (RT) in the management of NPC has been well established for many years. The influence of other parameters, for example, combination chemotherapy (CT) and the true extent of the primary lesion on survival rates have been, however, less clearly demonstrated in view of obvious inconsistencies in clinical data on post-CT treatment results (22,25) and the divergence of currently proposed staging systems (1, 3, 9, 23). In our current study, 17 parameters were evaluated separately first and then followed by stepwise multivariate regression analysis to determine a final set of independent prognostic factors that influence treatment outcome.
Presented at the 17th International
AND
MATERIALS
From March 1979 to December 1985,56 1 patients with epithelial malignancy of the nasopharynx were referred to our hospital for megavoltage RT with or without combination CT. Factors included in initial survival analyses are as follows: 1. Demographic factors: There were 407 males (M) and 154 females (F) with a M/F ratio of 2.64/l. Age (years) distribution ranged from 16 to 76 with a median value of 45. 2. Clinical factors: Primary tumor (T) and nodal (N) status were staged according to the TNM system of the American Joint Committee (AJC) (1). Since Tl and T2 lesions were difficult to separate, they were combined into one group. The tip of thyroid cartilage was arbitrarily used as an anatomic boundary to divide the location of regional nodes into upper and lower ones.
Congress of Radiology,
of Radiation Oncology, Chang Gung Memorial Hospital, #5, Fu-Shin St., Kwei-Shan, Tao-Yuan Hsien, Taiwan. Acknowledgments-The authors thank Chinying Chen Wang for her excellent assistance in statistical analysis. Accepted for publication 24 May 1990.
l-8 July, 1989, Paris, France.
* Department of Radiation Oncology. + Department of Internal Medicine, Division of HematologyOncology. Reprint requests to: Simon G. Tang, M.D., M.S., Department 1143
November 1990, Volume 19, Number 5
1. J. Radiation Oncology 0 Biology 0 Physics
1144
disappearance, ~50% reduction, or ~50% reduction, respectively, of all measurable disease at nasopharynx and regional neck nodes. 6. Types of relapse: Four hundred fifteen patients were free from any relapse. There were, however, 41 patients with primary (T), 15 patients with nodal (N), and 7 1 patients with distant (M) relapses only. The number of patients with T + N (N = 3), T + M (N = 7) N + M (N = 8), and T + N + M (N = 1) relapses were too small to be included in survival analyses.
3. Diagnostic factors: The majority of our patients (N = 507) had their nasopharynx biopsied to obtain a tissue diagnosis. Relatively few patients (N = 54) also had additional nodal excision to confirm metastatic status of nasopharyngeal origin. All previous histological reports were retrospectively reviewed by one pathologist and then classified into three types according to criteria proposed by the World Health Organization (WHO) (19). 4. Therapeutic factors: All patients were irradiated through two lateral opposed portals encompassing nasopharynx, oropharynx, base of skull, posterior choana, subdigastric and submandibular lymph nodes. The lower cervical lymphatic-draining sites were irradiated by one single anterior portal. Planned curative tumor doses via external beam were 64.8 to 70.2 Gray (Gy) in 36 to 39 fractions, 7 to 8 weeks; this was then followed by intracavitary insertion of a high dose rate 6oCo source into nasopharynx by means of the Remote After-Loading (RAL) technique to deliver 10 to 16.5 Gy boost dose in 2 to 3 fractions, spaced 1 to 2 week(s) apart. Cervical lymphatic sites received 46.8 Gy at a depth of 3 cm which was then boosted to grossly involved nodes for an additional 20 to 30 Gy with 9 to 12 MeV electron beams. Patients with T4 and/or N3 lesions might have received two courses of chemotherapy with cis-diamminedichloroplatinum (CDDP, 100 mg/m2/day, IV infusion on day 1) and 5-fluorouracil (5-FU, 1000 mg/m2/day from day 2 to day 6, constant infusion) prior to a planned courses of RT. 5. Response rates at primary and nodal sites: Complete, partial, or no responses (CR, PR, NR), that is, complete
Each possible prognostic factor was first evaluated by univariate survival analyses using the actuarial or “life table” methods (5), whereas a generalized Wilcoxon test (7) was used to compare the significance of differences between paired survival curves. Cox’s proportional hazard model (4) using BMDP software package (6) was then used to test the joint effects of all previously screened significant factors. The median follow-up period was 20 months. RESULTS
As shown in Figure 1, the only demographic factor found to be of statistical significance was age at diagnosis. Patients less than or equal to 40 years of age had a better survival than those over 40 years (p = 0.0001). On the other hand, all of the following listed clinical factors correlated well with survival, that is, asymptomatic patients, patients with negative neck nodes or distant metastases, patients with Tl-T2 disease, Nl or upper neck node leB 1.0’ MOMTOO-
.mQ. .eO-
b
.AO-
.
1 b .aOo-
.ZO-
.lO-
0.01
0.0
4
0.0
8.0
24
1s
Months a
WI
335
b
I54
144
ml
yt)
I
a!2
40
40
do
84 8
0.0 ’
a:0
ir
?? ftw tradnmt
N
ah
40
4s
m
84
Monthsotto7t7o8hnwl
n
1,”
72
42
m
I7
166
1%
118
69
44
31
20
13
5
395
373
240
Fig. 1. Survival curves by demographic factors: (A) Sex; of patients at risk at various time intervals.
(J6 l-,5
55
40
25
I7
7
%3
63
37
26
15
(B) Age in years. Bottom column indicates the number
B ID .w
.M) .M 40
I .ao-
30
m-
.20
.10-
e: Eons Marrow (NH) f : lnguinal node(a) &J&) g: Axlllary node(s) (t&6) PCO.05
-‘.
Fig. 2. Survival curves by clinical factors: (A) Symptoms; (B) T classification; (C) N classification; (D) Nodal location: (E) Metastatic sites.
e
1-L b
d
i4
Months
j2
io
after treatment
48
$6
64
-33 pW3!lUa~aJJ!pU~ 33 'g!pUn wUoup.xe3 1[a3snowenbs palylua~agpdpood %U!Z!U!~I?.I~~-UON :-J-JSa-d XN paie’lua~a~rp-Klal~apour %u!z!ugmay-uoN :33fj a-m ~>IN 3xuou~m3 IIa3 snounmbs WDOu~3lES Ila3snowenbs %u!z!u!wax :33s a-M XI) K~o~oyledols~H (v) :s~opq 3you%!p 6q saNn3 [eA+tns 'E %a
pawgua.xagp-IlaM
.
V
ssa]‘aIdmexa _~oj‘sqwap Jayea u! paqnsal salIs~uels~p
~0 ~J!M luawaAloAu! atiau pz!rue13 OIU! pljo uMopyeaJq
q~IMsluarledo]pa-eduIo~sa~w leA$uns.n?alC-f %()z ueq$
qn3gyp aq ‘JaAaMoq ‘h.u uoyse~u! IInysleseq Inoql!M
uaaunod
pin03 p+rhIns pm uoymj!sse~3 1 uaamlaq uopelauo3 ou aj]oM pm Jayea jo sapas aq$ ur ‘$~j UT‘xuhqdoseu JO esomzu ayl qleauaq suorsal %ugo~yyu! JO luaixa anIl aql %upeauqap u! tualqold aql isaNal pIno s!q~ '(HZ %J)
lnoj ‘lapouruo~ssa;_l%a~ss,xo3 OIU!paJa$ua alaMslowpaId asay% uaqM
3asQeue
alegA!un
%uIsn IeAyns
JO ~101
-3FpaJdluwjyk?ys Ll~esgsye~saq 01 punoj a.wM saIqepeA '(gyo > d)sasdeIa~ou .IO‘p?pou ‘Lnzuyd @IO
UI aDuaIaj.j;rp ejo aA!wpald os~ealaM asdelaljosadly
uo wzdtu! aJ!uyap e alley suoymj!sse~3 N pue 1 ayL ‘(t767/0 = d) lueD,y!u -%!s@yspe~s IOU SBM acmaJaJ!p aql ‘p+4.ms JO suual u! waged aIm,u ueq~ lauaq op 01 waas sluaped alew -aJ ltq awrpu! elep Ino q%noqqv '(sl‘~~)says lay10 rlq paunyuo3 uaaq seq (pro s.n?aLopr) sluagvd %unok
1~ asdelaJ'9 a_nGg t.uo~j uaas aq ue3 sv 'satin3IeAiA\"ns
laqvnj e 10 pL pUE7 EL uaarnlaquogwedas
pJEMO$
sysou%oJd alqIxoAt?j aql ‘a%
v *lEA!AInS
'(SO'0> d)sluayWd XN .IOad ueq$ aler IeA!A.Ins laq%!q e pap!Aold sluayd aaq-apou pz 1 snld dno.C4aIqmomj srryl s$uaged pawa 111~ JO ?&s .IO%,SL.~oj palunom? sa$!s lepou Ivuo~a~ .IOLnmpd aqt Ia ~3 %ururelqo wa[qns leql sawlsuowap g a.m%y . . ‘(g)‘o > d ‘av ‘%LJ)Ldwaq)om
01 I3adsa.t ql!M
-aq3~ueAnfoaujosas.rno3o~i~uahuapunoqMs~uay3d lyauaq IeA!lunsaAgISOd
‘~!qd~~~ouIapjoQapeA~~ojI~ArrhlnsIeIJen)~~o~suo!~nq
~oj an.I$SBM al!soddo aql SeaJaqM ‘(g)'o> d ‘3~ ‘%+J)
%!sdeIal pue‘aA~suodsa.~‘~y~adwaq~‘zysou%~p‘~e~ur~~ . . -yuo3 luapuadapur aql alenleha01 pau%pap SBM aAp3adsoqa~lua.un3~noQl jo aoi~j
~jo
snws
$uexj$3!s
Ma!rl\aJ
‘11 ‘z)a~nl~~a~rIaq~u!~dN
cysou%o_Idjo saypnlsIelaAas
01 ~SE.I~UO~ UI
'(80'0 = d)auyap~oq uoywxj~sse~3
~om?j luapuadapu! -lea_Il01 loud sa]qepeA
a3uexj!u%!s unwodtu~
aql sealaqm ls01.11 a@s
01
Quo ‘(~00~0
aqlse~
cyx?wyaliy '(1 alqel) Quapuadapu!
set
> d) Juaw
pauyqo
aJaM
JO las ~euy e ‘snqL
.sas&?ue av+n?Agpnu aq) UT lyauaq zgsou%oJd qaql ISOI ‘-~vajo uoypp~? aql put? ‘sapou @uo!%aJ .IOxuhJeqdos -eu 01 sasop uoyzfp~~ ‘uoyz3o~ ~epou ‘s!wql ‘salqy.n?A
I? papIa!6 ICd~.Iaq$Lq3wq~vx
JO uog~ppe aql -(g~)o.o= d)
salis IBPOU~euo~%~01 AD
8’99 JO (SO'0 > d) alisheurpd aqlO$ &I 8379zjO aSOp uo!~e1pelepaA!a3a~s~ua1~ed uaqMuaasse~~gauaq3ysou -%old aq~‘8~pue vp sa&k~ UT uMoqs sy *p3+uns ql!M IIaM pa%t.?-;-;-;-;-;-;-;-;-;-;-;-;-;-;-;-;-;-;sa1qy.m cgnadmaql moj 111~ jo sn$e$S ‘(~90’0 = d ‘8f ‘%!J) uo!$Wj!SSEI3@3oIoqled S&OHM JO sad& snop_eA aq) qyM palou SBM a3ualajyp p3+uns ON '(g)'() > d ‘Vf %.J dnoJ% lawoj 01 p.lEMOJlyauaq p+4ms e SBM alaqL ~uop~3xa ~epou pzuopal pwog!p -pe Inoq$!m auoIe Lsdorq pza%hJeqdoseu e ql!M pamnj -uo3 aseasrp peq sluaped mo JO oh&j Llawu!xo~ddv ‘(gyo > d ‘z *%LJ)suo!saI y3aU _IaMOI .I0 EN-ZN ‘PlL-fL qJ!M s1uagcllllllllllllllllll.l.l.l.l.l.l.l.l D,!Jt?JS -clam ‘zym.uo$dmLs myI IaSJuol Ague~y!u%!s paA!1suo!s apOU
b
M
00
0.0
,,
i
I,
1,
28
kks’hs alla
12
40
5*
00
8.2
1e
24
wmms
treernent
32
10
48
Fig. 4. Survival curves by therapeutic factors: (A) external radiation dose to nasopharynx dose to nodal sites (C) Remote After-Loading brachytherapy (D) chemotherapy. A
58
04
after treamant
(B) external
radiation
B
1.0
--.
30
.a0
.TO
.eo
30
.40
30
.30-
.20
,
0.0
8.0
,e
24 Month
52 after
40
42
64
’
M
teatmnt
Fig. 5. Survival curves by response rates at (CR = Complete Not applicable = Patients with no neck nodes) (A) Primary
i,’
Month8
attw
treatment
response; PR = Partial response; sites; (B) Regional nodal sites.
NR = NO response.
1. J. Radiation Oncology 0 Biology 0 Physics
November 1990, Volume 19, Number 5
have demonstrated, however, that the leading prognostic value of M status has no dependency on any confounding factors under Cox’s proportional hazard regression analyses (Table 1). In Neel’s series ( 16) there was a definite correlation between various WHO’s pathologic types and treatment results. In contrast to this finding, however, we excluded the pathologic variable from the multivariate analysis because we were unable to demonstrate its prognostic significance from currently available data. Survival curves by therapeutic factors (Fig. 4) revealed that a dose-survival relationship existed, that is, a difference in survival as a result of (a) ~64.8 Gy vs 264.8 Gy to the nasopharynx, (b) ~66.8 Gy vs 266.8 Gy to regional nodal sites, and (c) “yes” vs “no” brachytherapy dose to the primary sites. The lack of this relationship after multivariate analyses indicated that radiation dose depends on other covariates to yield its prognostic value. In our clinical experience, a higher radiation dose often leads to higher response rates and this may, in turn, result in a better survival compared to lower doses. The role of CT remains controversial. Improved survival with combined RT + CT vs RT alone was reported by some series (14) whereas’ others (20, 2 1) could not confirm its positive value. In our series, only high risk patients with T4 and/or N3 lesions underwent neoadjuvant CT. Thus, one may attribute the poor outcome in the RT + CT group to an obvious selection bias. This explanation, however, does not hold true as one realizes that the CT variable was ranked 5th in prognostic value independent of T, N and M classification. According to our clinical experience, CT may cause an initial reduction of tumor burden in the primary and nodal sites. This, however, has not translated into any improvement in ultimate tumor control as dictated by survival. A possible interpretation of the adverse effect of pre-RT CT is that (a) neoadjuvant CT may unduly postpone the subsequent standard treatment with RT (b) patients may data
0.0 0.0
S:O
1'0
$4
2-2 4-O 48 treatment
60
04
Mdntho afler
Fig. 6. Survival curves by types of relapse (T = Primary relapse; N = Nodal relapse; M = Distant relapse; No = No relapse).
be identified (2). The location of cervical lymph node metastases, on the other hand, related closely to the prognosis as originally proposed in Ho’s staging system (9). Unfortunately, when one attempts to adjust other prognostic factors simultaneously, this parameter lost its independent predicting value on survival. The status of distant metastases (M) has long been recognized as an important prognosticator in the AJC’s staging system. Although a positive finding of distant metastases forecasts a poor outcome, there was no evidence to suggest any survival difference among various types of metastases, for example, lung, liver or bone, etc. (Fig. 2E). Some investigators (12, 17) have related the incidence of distant metastases to the extent of N classification. Our
Table 1. Survival analysis with covariates-Cox Setup no. 1 2 3 4 5 6 7 8 9 10
Variable
entered
Metastatic status Primary response rates Types of relapse Age Chemotherapy Nodal response rates Symptoms Biopsy methods N classification T classification
* Degree of freedom. Note: Summary of Stepwise results.
DF*
1
Log likelihood
Regression
Improvement chi-square
Model
p value
Global chi-square
p value
2
-1519.241 - 1500.558
64.550 37.367
0.000 0.000
113.061 168.491
0.000 0.000
3 4 5 6
- 1489.320 -1477.218 -1471.705 - 1467.863
22.476 24.204 7.685
0.000 0.000 0.001 0.006
190.406 214.251 220.87 1 232.929
0.000 0.000 0.000 0.000
7 8 9 10
- 1467.440 -1461.578 - 1459.544 - 1458.007
6.845 5.724 4.068 3.074
0.009 0.017 0.044 0.080
239.291 245.448 253.563 254.619
0.000 0.000 0.000 0.000
11.024
Prognosis of NPC 0 S. G. J. TANG et (I!.
become reluctant to receive RT as a result of either a false sense of security for cure of cancer after CT-induced tumor regression or an uncomfortable experience in drug related toxicities such as stomatitis, nausea, vomiting, and diarrhea. In view of an apparent lack of improvement in survival for CT treated patients, we have recently conducted a randomized trial to compare the relative efficacy between combined modality therapy and conventional treatment
1149
with RT alone for patients with advanced NPC. In addition, a study of EB-viral capsid antigen specific antibodies, for example, IgG and IgA, has been currently in progress. Hopefully, the data generated from this potential prognosticator may help reach solid conclusions on the determinants of survival or other therapeutic endpoints. Moreover, independent prognostic factors identified in this series should be accounted for separately via stratification in all future clinical trials.
REFERENCES 1. American Joint Committee on Cancer. Manual for staging of cancer. Chicago: American Joint Committee; 1983:3233. Baker, S. R.; Wolfe, R. A. Prognostic factors of nasopharyngeal malignancy. Cancer 49: 163- 169; 1982. Chen, K. Y.; Fletcher, G. H. Malignant tumors of the nasopharynx. Radiology 99: 165- 17 1; 197 1. Cox, D. R. Regression models and life tables. J. R. Stat. Sot. 34: 187-202; 1972. Cutler, S. J.; Ederer, F. Maximum utilization of the life table method in analyzing survival. Ann. Life Ins. Med. II: 9-24; 1964. 6. Dixon, W. J.; Brown, M. B.; Engleman, L. Statistical software. Berkeley: University of California Press; 198 1. 7. Gehan, E. A. A generalized Wilcoxon test for comparing arbitrarily single-censored samples. Biometrika 52:203-223; 1965. carcinoma: present status 8. Henderson, B. E. Nasopharyngeal of knowledge. Cancer Res. 34: 1187- 1188; 1974. 9. Ho, J. H. C. The natural history and treatment of nasopharyngeal carcinoma. In: Lee-Clark, R., Cumley, R. W., McCay, J. E., Copeland, M., eds. Proceeding of the X International Cancer Congress, Vol. 4. Chicago: Yearbook Medical Publishers; 1970: l- 14. and clinical study of naso10. Ho, J. H. C. An epidemiologic pharyngeal carcinoma. Int. J. Radiat. Oncol. Biol. Phys. 4: 183-198; 1978. Il. Hsu, M. M.; Huang, S. S.; Lynn, T. C.; Hsieh, T.; Tu, S. M. The survival of patients with nasopharyngeal carcinoma. Otolaryngol. Head Neck Surg. 90:285-295; 1982. carcinoma in Tai12. Hsu, M. M.; Tu, S. M. Nasopharyngeal wan. Clinical manifestations and results of therapy. Cancer 52:362-368; 1983. Carcinoma: A review of 1605 13. Huang, S. C. Nasopharyngeal patients treated radically with Cobalt 60. Int. J. Radiat. Oncol. Biol. Phys. 6:401-407; 1980. 14. Huang, S. C.: Lui, L. T.; Lynn, T. C. Nasopharyngeal cancer: Study III. A review of 1206 patients treated with combined modalities. Int. J. Radiat. Oncol. Biol. Phys. 11:1789-1793; 1985.
15. Levendag, P. C.; Huygen, P. L. M.; Kazem, 1.; Broek, P. V. D.; Slooff, J. L. Malignant tumors of the nasopharynx: retrospective review of 74 cases. Clin. Radiol. 34:45 l-458; 1983. 16. Neel, H. B. Nasopharyngeal carcinoma. Clinical presentation, diagnosis, treatment and prognosis. Otolaryngol. Clin. N. Am. 18(3):479-490; 1985. 17. Petrovich, Z.; Cox, J. D.; Middleton, R.; Ohanian, M.; Paig, C.; Jepson, J. Advanced carcinoma of the nasopharynx. 2 pattern of failure in 256 patients. Radiother. Oncol. 4: 1520: 1985. 18. Qin, D.; Hu, Y.; Yan, J.; Xu, G.; Gai, W. Analysis of 1379 patients with nasopharyngeal carcinoma treated by radiation. Cancer 61:1117-l 124; 1988. 19. Shanmugaratnam, K.; Sobin, L. Histologic typing of upper respiratory tract tumors, In: International histological typing of tumors. Geneva: World Health Organization; 1978. 20. Tannock, I. F.; Browman, G. Lack of evidence for a role of chemotherapy in the routine management of locally advanced head and neck cancer. J. Clin. Oncol. 4: 112 1- 1126: 1986. 21. Tes, P.; Ho, J. H. C.; Choy, D.; Choi, P.; Tsui, K. H. Adjunctive chemotherapy to radical radiation therapy in the treatment of advanced nasopharyngeal carcinoma. Int. J. Radiat. Oncol. Biol. Phys. 13:679-685; 1987. 22. Tsujii, H.; Kamada, T.; Takamura, A.; Matsuoka, Y.; Usubuchi, H.; Irie, G. Improved results in the treatment of nasopharyngeal carcinoma using combined radiotherapy and chemotherapy. Cancer 63:1668-1672; 1989. 23. Union International Contra Cancer. TNM classification of malignant tumors, 2nd edition. Geneva: UICC; 1974:3031. 24. Yeh, S. Some geographic aspects of most common diseases in Taiwan, part two. Int. Pathol. 7:24-28; 1966. 25. Zhang, E. P.; Lian, P. G.; Cai, K. L.; Chen, Y. F.; Cai, M. D.; Zheng, X. F. and Guang, X. X. Radiation therapy of nasopharyngeal carcinoma: prognostic factors based on a IO-year follow-up of 1302 patients. Int. J. Radiat. Oncol. Biol. Phys. 16:301-305; 1988.