Path. Res. Pract. 188,510-516 (1992)

Prognostic Factors in Squamous Cell Carcinoma of the Oral Cavity A Retrospective Study of 80 Cases

c. A.

Beltrami, L. Desinan and C. Rubinj1

Institute of Pathological Anatomy, University of Udine, Italy; llnstitute of Pathological Anatomy, University of Ancona, Italy

SUMMARY

A multiparameter analysis of 80 squamous cell carcinomas of the oral cavity was performed to determine the relative prognostic weight of the location of the tumor, tumor size (T), microscopic grade (G) and DNA content measured by means of flow cytometry. Tumors of the lip have an invariably favourable outcome, while at the other locations they have high rates of mortality (total 5 years of survival less than 35 %). A simple statistical analysis was performed using the concepts of death sensitivity (DS), death specificity (DSp), life sensitivty (LS) and life specificity (LSp): the best measure of unfavourable outcome was represented by G3 (DS=0.69; DSp=l) and by DNA aneuploidy (DS=0.76; DSp=0.45), while the best measure of favourable outcome was represented by G1 (LS=0.53; LSp=O.96), T1 (LS=0.81; LSp=0.78) and by DNA diploidy of the tumor cells (LS=0.45; LSp=0.86). A survival analysis using a step by step regression model according to Cox was carried out in order to evaluate more precisely the relative importance of prognostic factors: traditional prognostic factors (histological grade=G and tumor size=T) showed a strong statistical significance, while DNA content added some additional prognostic information only if associated to the macroscopic features of the tumor.

Introduction

Oral cancer includes malignant tumors of the oral and oropharyngeal mucosa, together with those of the tongue and lips. As all cases we studied were squamous cell carcinomas, we can consider them as a homogeneous group of neoplasias (Oral Squamous Cell Carcinoma: OSCC). The traditional prognostic factors are considered to be stage of disease, location, and histologic grade; a better prognosis is significatively related to a limited stage of disease or a low grade:" Moreover, the site of the neoplastic growth is tightly linked to prognosis: in fact, intra-oral 0344-0338/92/0188-0510$3.5 0/0

cancer has a much poorer prognosis than lip cancer and survival rates decline progressively according to tumor location from the lips, to the tongue, floor of the mouth, gingiva and palateS, 9, 11, 18, 19,20. The determination of DNA content using cytometric techniques (static and flow cytometry) is under investigation to evaluate its prognostic value 14,15,21,22. Franzen et al 8 published a preliminary study on 39 squamous cell carcinomas of the lip, oral cavity, oropharynx, and larynx using static cytometry on suspension of cells obtained from formalin-fixed, paraffin-embedded squamous cell carcinomas: 18 cases were DNA diploid or near-diploid; 15 DNA polyploid, and 6 DNA aneuploid. © 1992 by Gustav Fischer Verlag, Stuttgart

Prognostic Factors in Oral Carcinoma· 511

Successively, Tytor et aF3 performed a cytofluorimetric DNA analysis of 88 squamous cell carcinomas of the oral cavity. 48 % (42/88) ofthe tumors were DNA non-diploid. A correlation existed between the percentage of DNA non-diploid cells and increasing size of the tumor, decrease in histological grade, and presence of lymph node metastases; the S-phase level seemed to increase in less differentiated tumors, but did not correlate either with tumour size or with duration of symptoms. Moreover, the estimation carried out of the S-phase in DNA non-diploid tumors was difficult to evaluate. Kokal et aP6 studied 76 patients with primary resectable squamous cell carcinomas of the head and neck region: Cox regression analysis demonstrated that tumor DNA content was independent of clinicopathologic features and was the single most important prognostic factor in predicting relapse and death. Ensley et aF evaluated the presence and degree of DNA aneuploidy as measured by DNA index (DI) and the S-phase fraction (SPF) by means of flow cytometry in 294 specimens from 237 patients with untreated and recurrent squamous cell carcinomas of the head and neck. Tumor grade was compared to DI and SPF: the conclusion was that no relationship was present. Comparison of untreated and recurrent specimens suggested that portions of these tumors displaying DNA aneuploidy may be more vulnerable to cytotoxic therapy than diploid tumors. This phenomenon was referred to a low number of cells undergoing DNA synthesis, or to undefined characteristics, independent of kinetic considerations, conferring pharmacological resistance to the cells. The aim of our study is to evaluate the different weight of the various prognostic parameters using an adaptation of the statistical concepts of sensibility and specificity for diagnostic tests. In addition, for a better evaluation of results, we carried out a multivariate statistical analysis using the Cox model in order to analyze the influence of synchronous consideration of the various parameters. Material and Methods The population under study consisted of 80 patients with squamous cell carcinomas of the oral cavity. 73 (91 %) were male, aged 32-84 years (mean 61), and 7 (9 %) were female, aged 39-84 (mean 69). The clinical characteristics recorded in all the patients included age, sex, site of the primary tumor, stage, therapy after biopsy, smoke and alcohol consumption, oral hygiene and use of prosthesis, familiarity for the disease or for neoplasias at other sites and follow-up. All cases were classified according to WHO for histological diagnosis, grading (G) and pathological staging (pTNM). We accepted the recommendations for the malignancy grading of Anneroth et aU and the grading was performed within the histologically most invasive areas of the tumors4 . DNA measurements: 50 [tm thick sections were prepared from representative tumor areas from formalin-fixed, paraffin-embedded material using the method described by Hedley l2 with some modifications. In brief, sections were cut from the paraffin blocks, dewaxed in xylene and rehydrated in a series of decreasing

Table 1. From a statistical point of view, sensitivity is defined as the proportion of patients who have the disease in whom the test result is positive (this is a measure of the probability that a person who has the disease will have a positive test result), and specificity as the proportion of healthy subjects in whom the test result is negative (this measures the probability that a healthy person will have a negative test result). In this investigation, we assumed death sensitivity (DS) as the proportion of patients who died in whom the parameter under investigation (i.e. G, T, and DNA content) was present and death specificity (DSp) as the proportion of patients alive in whom the same parameter was absent; life sensitivity (LS) was defined as the proportion of patients alive in whom the parameter was present; and life specificity (LSp) as the proportion of patients dead due to disease in whom the parameters were absent. Parameters under analysis Present

Absent

Present

a

b

Absent

c

d

Death sensitivity Death specificity Life sensitivity Life specificity

(Ds) (Dsp) (LS) (Lsp)

a/(a+c) d/(b+d) a/(a+c) d/(b+d)

concentrations of ethanol. After washing, the tissue was resuspended in a 0.5 % pepsin solution for 60 min. Staining of the samples was done using ethidium bromide. The measurements of the DNA content were performed with an Epics C Flow Cytometer (Coulter Electronic, Hialeah, Fla) with the lise of the 488 nm (15 mw) band of an air cooled argon laser. At least 10,000 nuclei were analyzed for each case. Histograms were classified, according to Joensuu and Klemi!3 and Bronner et a12 with some modifications: 1) diploid (D), in the presence of one symmetrical GO/G1 peak; 2) peridiploid (P), in the presence of one asymmetrical and/or large GO-G1 peak; 3) aneuploid (A) if two GO/G1 peaks were present: in this case the first peak was considered diploid. Biparametric histograms (FS or SS against DNA fluorescence) were also performed: they are particularly useful for discriminating two populations with small differences in DNA content when the DNA histogram is peridiploid. Analysis of data: the prognostic significance of the two most important clinico-morphological parameters (tumor size = T, and histological grade = G) and of DNA ploidy was evaluated in a simple way using an adaptation of the concepts of sensitivity and specificity used for diagnostic tests (Table 1)3. Other parameters were also analyzed (age, macroscopic appearance, site, smoke, recurrences, sex, lymph node metastases) using the step by step regression model of Cox in order to evaluate the importance of groups of parameters. The cumulative survival rate was estimated with the product-limit method of Kaplan-Meier, and the comparison of the cumulative survival rate between groups was carried out with the generalized WilcoxonBreslow and Mantel-Cox tests. Cox's proportional hazard model was used to assess the relative importance of prognostic factors. The multivariate analysis was preceded by a study of the single factors in order to verify the most effective way to combine various subgroups. All p-values were nvo-tailed.

512 . C. A. Beltrami, L. Desinan and C. Rubini

Results

(p = 0.0001).To determine the relative importance and independence of the factors that were important in a univariate analysis, Cox's stepwise multivariate analysis was performed. The results obtained confirm that the presence of at least one of the following variables: T3 and T4 and G3 are associated with a high significance (p = 0.0001) to a more unfavourable prognosis. As far as the other combinations of anatomo-clinical features are concerned, a statistically significant result was obtained only for ulcerated tumors with an aneuploid content of DNA (p = 0.04) (Table 4).

The cases were divided into groups according to their follow-up (Tables 2-3}. Patients alive with a short followup and patients deceased for reasons unrelated to the diseasewere not subjected to further investigation, because no prognostic information could be obtained. On the other hand, patients alive with a long follow-up and patients deceased as a result of the disease were subjected to statistical evaluation with the aim to weigh the importance of various prognostic factors. The results of the analysis of the cases using the concepts of death sensitivity (DS), death specificity (DSp), life sensitivity (LS) and life specificity (LSp) for the most important clinico-morphological parameters (T and G) and for DNA content were the following: G3 (DS=0.69; DSp=1; LS=1); G2 (DS=0.26; DSp=0.53; LS=0.47; LSp=0.73); G1 (DS=0.04; DSp=0.47; LS=0.53; LSp=O.96). T4 (DS=0.30; DSp=1; LS=O); T3 (DS=0.10; DSp=1; LS=O); T2 (DS=0.36; DSp=0.88; LS=O.12; LSp=0.63) T1 (DS=O.22; DSp=O.12; LS=0.81; LSp=O.78). DNA Aneuploidy (DS=0.76; DSp=0.45; LS=0.54; LSp=0.24) DNA Diploidy (DS=0.62; DSp=0.54; LS=0.45; LSp=0.86). The most powerful independent factors were histological grade (G) (~= 0.0001) and tumor size (T)

Discussion Most of the general characteristics of the population under investigation confirm previous reports 6, 10, 17: a larger predominance of the male sex (9: 1), the prevalence of the old age group (male 61 y, female 69 y) and the presence of two major risk factors, i.e. smoke and alcohol; the association with other neoplasias (particularly of the upper respiratory tract) was observed. The highly characteristic association with cancer of the larynx and of the lung probably reflects the same drinking and smoking habits of this population. Location of the primary tumor has a powerful prognostic significance: in fact, lip neoplasias have a better outcome than carcinomas arising in the oral cavity. There are two possible explanations for this phenomenon: a

Table 2. a) Patients alive with a short follow-up «5y); b) Patients alive with a long follow-up (>5y) S/Age

Site

Macr G

a) Short follow-up «5y) 1 M 61 trig veg 2 M 45 floor veg 3 M 66 pal ule

pTNM

DNA Rec

Fam

Smoke Aleohol Plate

Hygiene

Therapy tct+chem surg surg+tct +cherr surg+tct surg+tct surg+tct tct

17 20 44

surg surg surg-tct surg surg+tct surg+tct surg+tct surg surg surg surg+tct surg+tct surg+tct surg+tct surg tct surg

66 73 74 77 84 86 86 92 92 97 109 117 126 126 130 143 177

G2 G3 G3

T2N1MX Dipl T2NOMO Dipl T2NOMO

no no yes

yes no no

+++ +++ ++ ++ ++++ +++

no no no

scarce scarce

veg ule ule veg

G3 G3 G2 G2

T2NOMO TlNOMX TlNOMO T2NOMO

Aneu Dipl Dipl Dipl

no no no no

yes yes yes no

+++ + +++ +++

no yes yes no

good

b) Long follow up (>5y) veg 1 M62 lip 2 M 70 floor ule 3 M72 lip ule 4 M59 lip ule veg 5 M39 lip ule 6 M43 tongue 7 M62 lip veg ule 8 M52 lip 9 F 82 gmg ule veg 10 M57 lip 11 M68 tongue ule 12 M 76 pal veg ule 13 M67 floor 14 M 36 lip veg 15 M65 floor ule 16 M55 tongue .. ule 17 M50 lip ule

Gl G2 G2 G1 Gl G1 G1 G1 G2 G1 G2 G2 G1 G2 G2 G2 G1

TlNOMO TlNOMO TlNOMO TlNOMO TlNOMO TlNOMO TlNOMO TlNOMO T2NOMO TlNOMO TlNOMO T2N1MO TlNOMO TlNOMO TlNOMO TlNOMO TlNOMO

Aneu no yes no Aneu no no Aneu no no Aneu no Dipl yes Dipl no no Dipl no Aneu no Dipl no Aneu no Dipl no yes

no no no yes no yes no no no no no no no yes no no no

++

4 5 6 7

M57 M 71 M53 M 75

floor ging floor cheek

++ ++ ++

+ ++ +/++ ++ ++ ++ ++ +++ +++ ++ +++ ++ + +/+++ +++ +++ +++ +++ +++ +++ ++ ++ ++ ++ ++ ++++ +

yes no no no no yes no no yes no no no yes no no yes no

scarce

scarce good scarce good

F-UP{m)

45 48 49 53

Prognostic Factors in Oral Carcinoma . 513 Table 3. a) Patients deceased for reasons unrelated to the disease; b) Patients deceased as a result of the disease pTNM

DNA Rec

Fam

Smoke Alcohol Plate

a) Decease unrelated to the disease 1 M 69 lip veg Gl 2 M 81 lip veg Gl 3 M 68 cheek ule G3 4 M 61 lip veg G2 5 M 73 lip Gl veg 6 M 68 ging G3 veg 7 M 72 cheek G2 veg

TINOMO TlNOMO T2NOMO TlNIMX TlNOMO TlNOMO T2NOMO

Aneu no no Dipl no Aneu no Aneu no Aneu no Aneu no

yes yes yes no no no no

++ ++

b) Decease as a result of the disease 1 M 65 pal ule G3 2 M 84 floor ule G3 3 M 76 tongue veg G3 4 M 66 tongue veg G3 5 F 84 ging veg G2 G3 6 M 76 floor ule 7 M 55 tongue ule G3 8 M 59 tongue ule G3 9 M 60 trig veg G3 10 M 54 ging floorveg G3 11 M 72 floor veg G3 12 M 67 floor ule G3 13 M 77 tongue ule G3 14 M 72 pal ule G2 15 F 75 tongue veg G3 16 F 64 cheek ule G3 17 M 62 pal ule G3 18 M 82 tongue veg Gl 19 M 33 tongue ule G3 20 M 39 cheek veg G3 21 M 52 floor ule G3 22 M 66 pal ule G3 23 M 50 tongue ule G3 24 M 49 cheek ule G3 25 M 79 pal veg G2 26 M 72 floor ule G2 27 M 57 pal veg G3 28 M 50 tongue veg G3 29 F 39 tongue veg G2 G3 30 M 53 tongue ule 31 M 53 cheek ule G2 32 F 76 tongue veg G3 33 M 71 tongue ule G3 34 M 63 floor veg G3 35 M 64 floor veg G3 36 F 64 cheek ule G3 37 M 48 tongue ule G2 ule G3 38 M 73 cheek 39 M 64 pal veg G3 40 M 65 pal ule G2 41 M 62 cheek veg G3 42 M 65 tongue veg G2 43 M 32 floor veg G3 44 M 62 pal veg G3 45 M 59 floor ule G2 46 M 65 pal ule G2 ule Gl 47 M 50 pal 48 M 59 tongue ule.G2 49 M 44 floor ule G2

T4NXMX T4NIMX T4NXMX T2NIMI T4NIMO T4NIMO T4NIMO T2N2MO T4NIMX T4NIMI T2NXMl T4NOMO T2N2MO T3NIMX T3NIMO T2NIMO TlNOMX T2NIMX T2NIMX TlNOMX T4NIMX T4NIMX TlNXMX T4NOMO TlNXMX T3NOMX T4NIMX T4NIMX TlNOMX T2NIMI T3NXMX T2NIMO T4NXMl T2NOMX T2NIMO T2NIMO T2NXMX TINXMX T3NIMX T2NIMX T2NOMX T2NOMO TlNOMO TlNOMX T2NOMO TlNOMO GINOMO TlNOMO T2NOMO

Aneu Aneu Dipl Dipl Aneu

no yes no no yes yes yes yes no no yes yes no no yes no no yes no no no no yes no yes no yes yes no no yes no yes yes yes no no yes yes no yes yes no yes yes no yes yes no

StAge

Site

Macr G

Aneu Aneu Aneu Aneu Aneu Aneu Aneu Aneu Aneu Dipl Dipl Aneu Aneu Aneu Aneu Aneu Dipl Aneu Aneu Aneu Aneu Aneu Aneu Dipl Aneu Aneu Aneu Aneu Dipl Aneu Dipl Dipl Dipl Aneu Aneu Aneu Aneu Aneu

no no no no no no yes no no no no no no no no no yes yes no no no no yes yes no no no no no no yes no no no yes yes yes no no yes yes yes yes no yes no yes yes yes

+/+ ++ ++ ++ ++

yes

scarce

scarce

+++

+

no yes yes yes yes no

+++ +++

+++ +++

++

yes

+++ +++

+++

+++

+ ++

++

++

+ +

++

+++ +++ +++ +++ +++

+++ +++

++ ++

+++

+

+

++

+++ +++

++

+++ +++ +++

++ ++

+++ +++ +++ +++ +++ +++ +/+++ +++ +++ +++ +++ +++ +++ +++ +++ +++

++,J-

+++

++

+++

++

Hygiene

++

+++ +++ +++

+

+++ +++

+ ++

+++

++ ++

+/-

++

+++

++ ++ ++

+++

++

+++ +++

++ + + ++

+++ +++

scarce

yes yes no yes no no yes no yes yes yes no yes yes no no no yes no no no no yes no no no no no yes no yes yes yes no no yes no no yes yes no yes yes no no no

scarce scarce good scarce scarce good scarce scarce

scarce worst scarce worst scarce scarce scarce scarce scarce worst

scarce scarce scarce

scarce scarce scarce

Therapy

F-UP(m)

surg+tct surg surg surg surg surg+tct surg

12 12 44 61 82 84 95

tct tct surg no surg tct surg+tct surg+tct tct tct chem adiuvant tct tct

3 3 3 3 5 6 6 6 7 7 8 8 8 8

tet

tct surg+tct surg+tct chem surg+tct tct tct surg+tct surg+tct surg+tct tct tct+chem tct+chem surg tct+chem surg+tct surg tct surg surg+chem surg+tct surg+tct surg+tct tct tct surg surg+tct surg+tct tct surg+tct surg surg+tct surg+tct surg

9

9 10 10 11 11 11 11 12 13 14 14 15 15 15 15 16 16 19 20 22 23 26 26 26 31 34 39 51 64 65 72 80 91 115

514 . C. A. Beltrami, L. Desinan and C. Rubini

The grading was always low (Gl or G2) in patients alive with a long follow-up. Patients deceased as a result of the disease showed a prevalence of higher grades (34 G3 cases, i.e. 69%). The diameter of the tumor, as revealed by T, was small in patients alive with a long follow-up (Tl and T2); patients deceased as a result of the disease showed a wide distribution of diameters from T1 to T4. The importance of pathological grade on the evolution of oral neoplasia is clearly demonstrated by the comparison of the two groups T1-alive and T1-dead: in this case, there were no statistical differences in the DNA content (euploid or aneuploid); however, the grading was G1 in the first group and G3 in the second. The prognostic significance of G and T, as shown by univariate analysis, is highly significant (p = 0.0001) (Fig. 1,2). The relation between DNA content and progression of neoplasia is more complex. The macroscopical features of the tumors (ulcerated or exophytic) were not prognostically significant (p = 0.1313), however, when

Table 4. Histological grade (G) and tumor size (T) were powerful prognostic factors. A statistically significant result was obtained for ulcerated tumors with aneuploid DNA content, while DNA content alone, and/or macroscopic aspect of tumor alone had no prognostic significance Parameter p>chi sq

=

G

T

DNA

Macro

Ulcerated tumor+DNA Aneuploidy

0.0001

0.0001

0.2256

0.1313

0.0419

greater possibility for precocious diagnosis evidenced by lower T and G of the tumors at this site, and/or a greater opportunity for a radical surgery. The site distribution of tumors of the oral cavity showed that the more difficult the inspection of the lesion was the more unfavourable was the prognosis: the whole group having a five-year survival of less than 35 %.

1.000 • • 0.900 0.600 S u

v Y

•

PARAMETER

0. • •'0 , I

=

G

•• Gl without LIP

0· G2

.0

,")q~

0.700

1.0.... -

I _

0.500

0,

...

0.400

0----. ••____•

0, 0,

iii

0.600

••



--O-Q

.Lr.

0.300

without LIP

•• G3 without LIP

'-'i. =

0.200 0.100

\ID-O-O

Ii...

0.000

o

20

---.

Fig. 1. Survival distribution of three risk groups according to histological grade (G) (lip location was excluded). Differences were statistically significant (p>chi sq = 0.0001).

.---+~--+----+----+---~

60

40

60

100

120

160

140

Follow up (mo)

PARAMETER

=

T

.-T 1

1.000 • •

0.900 0.800 S

u v Y

~

.>