VOLUME

25

NUMBER

24

AUGUST

20

2007

JOURNAL OF CLINICAL ONCOLOGY

O R I G I N A L

R E P O R T

Role of Human Papillomavirus Genotype in Prognosis of

Early-Stage Cervical Cancer Undergoing Primary Surgery
Chyong-Huey Lai, Chee-Jen Chang, Huei-Jean Huang, Swei Hsueh, Angel Chao, Jung-Erh Yang,
Cheng-Tao Lin, Shang-Lang Huang, Ji-Hong Hong, Hung-Hsueh Chou, Tzu-I Wu, Kuan-Gen Huang,
Chun-Chieh Wang, and Ting-Chang Chang
From the Departments of Obstetrics
and Gynecology, Pathology, and Radiation Oncology, Chang Gung Memorial
Hospital and Chang Gung University
College of Medicine; and Graduate
Institutes of Clinical Medical Sciences,
Chang Gung University, Taoyuan,
Taiwan.
Submitted February 16, 2007; accepted
May 21, 2007.
Supported by Grants No. NSC93-2314-B182-036 and NSC94-2314-B-182-011 from
the National Science CouncilTaiwan.
Presented in part at the 38th Annual
Meeting on Womens Cancer,
March 3-7, 2007, San Diego, CA.
C.-H.L., C.-J.C., and H.-J.H. contributed
equally to this work.
Authors disclosures of potential conflicts of interest and author contributions are found at the end of this
article.
Address reprint requests to ChyongHuey Lai, MD, Department of Obstetrics and Gynecology, Chang Gung
Memorial Hospital, 5 Fu-Shin St,
Kueishan, Taoyuan 333, Taiwan;
e-mail: sh46erry@ms6.hinet.net.
2007 by American Society of Clinical
Oncology
0732-183X/07/2524-3628/$20.00

Purpose
Our aim was to evaluate the prognostic significance of human papillomavirus (HPV) genotype
in early-stage cervical carcinoma primarily treated with surgery in a large tertiary referral
medical center.
Patients and Methods
Consecutive patients who underwent primary surgery for invasive cervical carcinoma of International Federation of Gynecology and Obstetrics (FIGO) stage I to IIA between 1993 and 2000 were
retrospectively reviewed. Polymerase chain reaction (PCR) using a general primer set followed by
reverse-blot detection of 38 types of HPV DNA in a single reaction was performed for genotyping.
E6 type-specific PCR was performed to validate multiple types.
Results
A total of 1,067 eligible patients were analyzed. HPV DNA sequences were detected in 95.1%
of the specimens, among which 9.6% contained multiple types. HPV 16 was detected in 63.8%
of the samples, and HPV 18 was detected in 16.5% of the samples. The median follow-up time of
surviving patients was 77 months. By multivariate analysis, FIGO stage, lymph node metastasis,
depth of cervical stromal invasion, grade of differentiation, and HPV 18 positivity were significantly
related to cancer relapse. FIGO stage II, deep stromal invasion, parametrial extension, HPV 18
positivity, and age older than 45 years were significant predictors for death. Using the seven
selected variables from either recurrence-free or overall survival analysis, death-predicting
(P .0001) and relapse-predicting (P .0001) models classifying three risk groups (low,
intermediate, and high risk) were constructed and endorsed by internal validation.
Conclusion
The independent prognostic value of HPV genotype is confirmed in this study. The prognostic
models could be useful in counseling patients and stratifying patients in future clinical trials.
J Clin Oncol 25:3628-3634. 2007 by American Society of Clinical Oncology

DOI: 10.1200/JCO.2007.11.2995

INTRODUCTION

Cervical cancer remains the second most prevalent

female malignancy in the world.1 Optimal management of cervical cancer consists of precisely evaluating tumor extent and implementing treatment
followed by deliberate post-therapy surveillance,
early detection of recurrence, and appropriate salvage therapy. For early-stage invasive cervical cancer, either radical surgery or radiotherapy (RT)
provides equal efficacy.1-3 Selecting appropriate
candidates for primary surgery and providing welldefined high-risk patients with postoperative adjuvant therapy is crucial.
Clinicopathologic factors for cervical cancer,
including International Federation of Gynecology
and Obstetrics (FIGO) stage, lymph node metasta3628

sis, parametrial or vaginal extension, nonsquamous

histology, grade of differentiation, lymphovascular
space invasion, and depth of cervical stromal invasion, have been noted to be associated with prognosis. Most studies are concordant regarding the
former two variables (stage and lymph node metastasis); however, selected significant prognostic covariates vary between studies according to the
patient population and methods of analysis.1-5
Although the molecular mechanisms of human papillomavirus (HPV) in cervical carcinogenesis have not been fully elucidated, HPV infection
has been established as a necessary cause for cervical
intraepithelial lesions and invasive carcinoma. With
adequately sensitive methods, the HPV DNA detection rates approach 95% to 100% in cervical carcinoma tissue.6-9 The relationship of HPV genotype to

HPV Genotype and Cervical Cancer Prognosis

prognosis of invasive cervical cancer has been controversial.9-23 Many

reports found that HPV 18 positive tumors were associated with
poorer prognosis.10-15 Some reported that serum antibody to HPV 16
E7 peptide,16 HPV 16 DNA positivity,17 and HPV 16/18 DNA positivity18 were associated with poor outcome, whereas HPV 16 DNA
positivity was related to better survival in another report.19 One report
found that HPV 58 related types were associated with better prognosis,20 whereas another study noted that HPV 31related types predicted better survival.21 In contrast, no prognostic value for HPV
status or HPV genotype was found by other studies.9,22,23
There are at least three reasons for the controversy. First, the
previous reports investigating the prognostic role of HPV genotype
are mostly small series (n 150).10,11,15,18,19,21,22 Usually, only casecontrol or univariate analyses were performed.9,10,18,19,23 Second, series able to perform multivariate analysis usually included cervical
cancer stages I to IV, where other variables could not compete with the
powerful factor of stage12,14,16,21 or lymph node metastasis.15 Third,
various methods of different sensitivity of detecting HPV were used.
Further studies with sufficient sample size and sensitivity in detecting
HPV DNA are necessary to draw conclusion.
The purpose of this retrospective study was to evaluate the prognostic significance of HPV genotype in FIGO stage I to IIA cervical
carcinoma primarily treated with surgery in a large tertiary referral
medical center. The target sample size was more than 1,000 patients.
PATIENTS AND METHODS
Study Population
A list of consecutive patients who underwent primary definitive surgery
or RT for invasive cervical carcinoma of FIGO stage I to IV between 1993 and
2000 at Chang Gung Memorial Hospital was retrieved from the hospital
database initially.24 Medical records were retrospectively reviewed. The institutional review board approved the study. Formalin-fixed paraffin-embedded
tissue specimens were used for DNA analysis. Patients who had a wrong
diagnosis or incomplete medical records, missing paraffin blocks, or specimens with inadequate DNA quality were excluded; the remaining patients
were eligible for the HPV genotype study.24 For prognosis analysis, only
patients with stage I to IIA disease receiving primary surgery were included in
the current report.
DNA Extraction, SPF1/GP6 Polymerase Chain Reaction,
and E6 Type-Specific Polymerase Chain Reaction
The procedures of DNA extraction and polymerase chain reaction
(PCR) have been detailed previously.24-26 Briefly, short PCR fragment 1
(SPF1)/general primer 6 (GP6)9,25 PCR was performed for 40 cycles. E6 typespecific PCR was performed for 50 cycles. The 25 sets of primer sequences and
procedures of type-specific PCR were detailed in previous reports.24 Routine
precaution procedures were applied to avoid carrying over or contamination.24-26
HPV Genotyping by Genechip
Fifteen microliters of the resultant PCR products were hybridized with
an Easychip HPV Blot (King Car, I-Lan, Taiwan; hereafter referred to as HPV
Blot) membrane in a single reaction. HPV Blot contains 38 types of HPV
(6, 11, 16, 18, 26, 31, 32, 33, 35, 37, 39, 42, 43, 44, 45, 51, 52, 53, 54, 55, 56, 58,
59, 61, 62, 66, 67, 68, 69, 70, 71 [CP8061], 72, 74, 81 [CP8304], 82 [MM4], 83
[MM7], 84 [MM8], L1AE5) oligonucleotide probes of 20- to 30-mer on a
nylon membrane. The hybridization and detection procedures were described previously.24-26
Validation of HPV-Negative Results and Multiple Types
Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) PCR was performed when HPV-negative results were obtained at the first round. If
GAPDH PCR was also negative, another block was used when appropriate. In
www.jco.org

case of GAPDH positivity and HPV negativity by both type-specific PCR of the
leading eight types (determined by the preliminary data) and a repeat of HPV
Blot, the final result was designated as HPV negative.24
E6 type-specific PCRs (25 types) were performed to validate multiple
types on HPV Blot. In case of discordance between the type-specific PCR
and HPV Blot results, a repeat of HPV Blot was performed to resolve
the discordance.24
Postoperative Adjuvant Therapy
Our policy of selecting patients for adjuvant therapy after primary surgery has evolved with time according to ongoing clinical trials. In general,
patients with positive vaginal margin, parametrial extension, and full thickness
of cervical stromal invasion would be offered adjuvant RT. For pelvic lymph
nodepositive patients, adjuvant chemotherapy, RT, or concurrent chemoradiotherapy (CCRT) was prescribed within or out of a clinical trial.27 The
regimen of chemotherapy and method of adjuvant RT have been described
previously.27,28 HPV genotype was not available for the decision of adjuvant
therapy during the study period.
Post-Therapy Surveillance
Our protocol of post-therapy surveillance consisted of follow-up visits
every 3 months for 2 years, every 4 months for the third year, every 6 months
between years 3 and 5, and yearly after 5 years. Clinical history, physical and
pelvic examination, Pap smear, and serum tumor markers were checked on
every visit. Yearly chest x-ray studies were advised in asymptomatic patients,
whereas computed tomography or magnetic resonance imaging scans were
performed yearly for the first 3 consecutive years for high-risk groups or when
clinically indicated.29
Statistical Analysis
The data were analyzed using the SPSS version 11.0 statistical package
(SPSS Inc, Chicago, IL). Pearsons 2 test was used to evaluate the association between covariates. Survival curves were generated using the KaplanMeier method. Multivariate analysis by Cox stepwise forward regression
was used for those covariates selected in univariate analyses of survival by
log-rank test with a P .05. Their hazard ratios (HRs) and 95% CIs were
calculated using the Wald test. Using the selected independent prognostic
variables, outcome-predicting models were formulated. Continuous variables (such as age) were determined by the recursive partitioning method
for optimal cut point analysis. Bootstrap resampling using R software
(http://www.r-project.org) was performed randomly to construct new
data sets (events per variable 10, n 1,067) followed by Cox regression
analysis, which was repeated 1,000 times, and the counts of selection as
significant were recorded for each variable.30 All tests were two sided, and
P .05 was considered statistically significant.

RESULTS

The hospital database identified 2,446 patients with FIGO stage I to IV

cervical cancer primarily treated with surgery or definitive RT/CCRT
between 1993 and 2000. Of the 2,446 patients, 161 with a wrong
diagnosis or incomplete medical records, 154 with missing paraffin
blocks, and 13 with GAPDH-negative specimens were excluded. A
total of 2,118 patients (1,072 receiving primary surgery and 1,046
receiving primary RT/CCRT) were analyzed for genotype distribution, the results of which were reported elsewhere.24 Of the 1,072
patients, five patients were subsequently excluded because of receiving
neoadjuvant chemotherapy (n 1) and primary RT (n 4) before
surgery; therefore, the remaining 1,067 patients were eligible for the
current study. The clinicopathologic characteristics of the study patients are listed in Table 1.
Surgical and Postoperative Treatment
All patients had a hysterectomy. The surgical procedures were
either simple or type 1/2 extended hysterectomy with (n 57) or
3629

Lai et al

Table 1. Characteristics of the Study Patients

Patients (N 1,067)
Characteristic
Age, years
Median
Range
35
35-50
50
FIGO stage
IA
IB
IIA
Histologic type
Squamous carcinoma
Adenocarcinoma
Adenosquamous carcinoma
Small-cell carcinoma
Grade of differentiation
Well
Moderate
Poor
Tumor size, cm
2
2-4
4
Lymph node metastasis
Yes
No
Depth of cervical stromal invasion
One third
One third to two thirds
Two thirds
Full thickness
Adjuvant treatment
No adjuvant
Chemotherapy
Radiotherapy
Concurrent chemoradiotherapy

No.

50
26-80
85
499
483

8
47
45

60
880
127

6
82
12

886
131
48
2
1,047
307
400
340
1,049
504
497
48
1,061
95
966
1,047
425
281
140
201

83
12
5

767
34
225
41

72
3
21
4

29
38
33
48
47
5
9
91
41
27
13
19

Abbreviation: FIGO, International Federation of Gynecology and Obstetrics.

Some of the histology slides were not available for scoring.

without (n 3) pelvic lymph node biopsy or sampling for stage IA

disease. Radical hysterectomy and pelvic lymphadenectomy (RHPLND; n 1,007) was performed for all patients who had a preoperative diagnosis of stage IB disease, and three stage IB patients had a
preoperative diagnosis of grade 3 cervical intraepithelial neoplasia and
underwent a simple hysterectomy followed by adjuvant RT. Three
hundred patients received postoperative adjuvant therapy (Table 1).
HPV DNA Analysis
Of the 1,067 eligible patients, HPV DNA sequences were detected
in 1,015 of the specimens (95.1%; 95% CI, 93.8% to 97.4%), among
which 90.4% harbored single-type and 9.6% contained multiple-type
HPV sequences. A total of 28 types of HPV were detectable in this
study. The leading 10 types were HPV 16 (63.8%), HPV 18 (16.5%),
HPV 58 (6.8%), HPV 52 (5.3%), HPV 33 (4.3%), HPV 45 (2.4%),
HPV 31 (1.1%), HPV 39 (0.8%), HPV 59 (0.6%), and HPV 53 (0.5%).
The other 18 types in order of decreasing frequency were HPV 67, 11,
43, 35, 82 (MM4), 51, 42, L1AE5, 70, 68, 37, 26, 81, 71, 72, 69, 62, and
3630

54. HPV 16 or 18 comprised 77% of specimens (822 of 1,067 specimens), and HPV 58, 33, or 52 was detected in 16.5% of specimens (176
of 1,067 specimens). Of the 1,015 HPV-positive specimens, 78 (7.7%)
contained double types, 18 (1.8%) contained triple types, and one
(0.1%) contained quadruple types.
Associations of HPV Parameters With Various
Clinicopathologic Characteristics
Associations of HPV parameters with various clinicopathologic characteristics are listed in Appendix Table A1 (online only).
By Pearsons 2 analysis, age, tumor size, FIGO stage, depth of
cervical stromal invasion, lymph node metastasis, and parametrial
extension were unrelated to HPV 18 positivity. HPV 18 positivity
was significantly more common in adeno-adenosquamous (43%)
than squamous carcinoma (11%; P .001) and presence of lymphatic permeation was also significantly associated with HPV 18
positivity (P .04). No variable was significantly related to HPV
status (positive or negative). HPV 58, 33, or 52 positivity was
significantly associated with age ( 35 v 35 years: 7% v 17%,
P .01; 45 v 45 years: 10% v 21%, P .001; 50 v 50 years:
11% v 23%, P .001) and histologic type (squamous v adenoadenosquamous: 18% v 7%, P .001).
Univariate and Multivariate Analyses of
Prognostic Factors
The median follow-up time of surviving patients was 77 months
(range, 0.1 to 161 months). Up to the date of analysis (December 30,
2006), there were 137 cancer recurrences and 152 deaths. Of the 137
patients who experienced a relapse, 94 died of disease, whereas the
remaining 43 were alive either without disease after salvage treatment
or alive with cancer. Of the 152 deaths, four patients who died of
intercurrent disease unrelated to cancer were censored at the date of
death for the calculation of cancer-specific overall survival (OS). The
other 40 patients whose cause of death on death certificates was cervical cancer without a definite recurrence record were counted as having
cancer-specific events. The remaining 14 patients with unknown
cause of death were censored.
Univariate analyses were performed using the log-rank test. Age,
stage, tumor size, grade, depth of cervical stromal invasion, lymphatic
permeation, lymph node metastasis, and HPV 18 positivity were significantly associated with both recurrence-free survival (RFS) and OS
rates, whereas histologic type, HPV status (positive or negative), and
HPV pattern (single type or multiple type) were unrelated to survival
(Table 2). Using an age cut point of 50 years, only FIGO stage II, deep
cervical stromal invasion, parametrial extension, and HPV 18 positivity were significant by multivariate analysis (Appendix Table A2, online only), whereas age was selected in addition to the former four
variables as independent predictors for OS using an age cut point of 45
years (Table 3). By multivariate analysis, FIGO stage, lymph node
metastasis, depth of cervical stromal invasion, grade of differentiation,
and HPV 18 positivity (P .009) were significantly related to RFS,
despite using different age cut points (Appendix Table A2 and Table 3).
Outcome-Predicting Models
Outcome-predicting models were constructed using the seven
significant covariates for either OS or RFS. The prognostic score
was 0 for FIGO stage of stage I and 1 for stage II; 0 for depth of
stromal invasion less than one third and 1 for stromal invasion
one third; 0 for no lymph node metastasis and 1 for lymph node
metastasis; 0 for no parametrial extension and 1 for presence of
JOURNAL OF CLINICAL ONCOLOGY

HPV Genotype and Cervical Cancer Prognosis

Table 2. Univariate Analyses of Clinicopathologic Covariates and HPV Parameters on Overall and Recurrence-Free Survival
Overall Survival
Variable
Age, years
45
45
50
50
FIGO stage
IA-IB
IIA
Tumor size, cm
2
2-4
4
Histology
Squamous
Adeno-adenosquamous
Grade of differentiation
1
2
3
Depth of cervical invasion
One third
One third to two thirds
Two thirds
Lymphatic permeation
No
Yes
LN metastasis
No
Yes
Parametrial extension
No
Yes
HPV status
Positive
Negative
HPV pattern
Single
Multiple
HPV 18
Positive
Negative
HPV 58, 33, or 52
Positive
Negative

No. of
Patients
1,067
411
656
584
483
1,067
940
127
1,049
504
497
48
1,065
886
179
1,047
307
400
340
1,047
425
281
341
1,055
805
250
1,061
966
95
1,067
1,007
47
1,067
1,015
52
1,015
918
97
1,067
176
891
1,067
176
891

5-Year Rate
(%)

93
90
92
90

Recurrence-Free Survival
P
.0001
.0001

5-Year Rate
(%)

92
87
90
88

.0001
92
83

.0008
.03
.0001

91
74
.0001

97
87
72

.0001
95
85
67

.16
92
86

.25
90
86

.0001
97
90
86

.0001
97
87
84

.0001
97
90
84

.0001
95
88
83

.0001
93
85

.0001
91
82

.0001
92
78

.0001
91
72

.0001
92
70

.0001
90
68

.61
91
88

.49
89
90

.91
91
92

.52
89
90

.01
85
92

.03
84
90

.53
94
91

.07
93
88

Abbreviations: HPV, human papillomavirus; FIGO, International Federation of Gynecology and Obstetrics; LN, lymph node.

parametrial extension; 0 for differentiation of grade 1 and 1 for

grade 2 or 3; 0 for HPV 18 negativity and 1 for HPV 18 positivity;
and 0 for age 45 years and 1 for age more than 45 years. The
probability of dying was highest in the high-risk group (prognostic
score 4 to 7; HR 7.8; 95% CI, 4.8 to 12.7), followed by the
intermediate-risk group (prognostic score 3; HR 5.0; 95% CI,
3.1 to 8.1), when compared with the low-risk group (prognostic
score 0 to 2; HR 1.0; P .0001; Table 4). Five-year OS rates of
the high-, intermediate-, and low-risk groups were 97.4%, 87.6%, and
77.9%, respectively (Fig 1A). The probability of cancer recurrence was
www.jco.org

highest in the high-risk group (prognostic score 4 to 7; HR 6.5;

95% CI, 4.1 to 10.1), followed by the intermediate-risk group (prognostic score 3; HR 3.4; 95% CI, 2.1 to 5.2), when compared with
the low-risk group (prognostic score 0 to 2; HR 1.0; P .0001;
Table 4). Five-year RFS rates of the high-, intermediate-, and low-risk
groups were 95.9%, 86.7%, and 72.7%, respectively (Fig 1B).
Internal Validation by Bootstrap Analyses
Internal validation using bootstrapping by resampling 1,000
times showed that the significant prognostic factors were the same as
3631

Lai et al

Table 3. Multivariate Analysis of Clinicopathologic Covariates and HPV Parameters on Overall and Recurrence-Free Survival
Overall Survival
Variable
Age: 45 v 45 years
FIGO stage: II v I
Tumor size: 4 v 2-4 v 2 cm
LN metastasis: yes v no
Grade
2v1
3v1
Depth of cervical stromal invasion
One third to two thirds v one
third
Two thirds v one third
Parametrial extension: yes v no
HPV 18: positive v negative

Recurrence-Free Survival

HR

95% CI

HR

95% CI

1.7
1.8

1.1 to 2.6
1.2 to 2.8

2.6

2.2

1.8 to 3.9

1.4 to 3.4

2.1
2.1

1.1 to 3.9
1.1 to 4.0

2.4

1.4 to 4.3

.02
.003
.15
.11
.42
.96
.41
.001
.002

2.0

1.1 to 3.5

.14
.001
.17
.001
.05
.02
.02
.03
.02

3.2
2.6
1.7

1.9 to 5.5
1.5 to 4.4
1.1 to 2.6

.001
.001
.01

2.0

1.8

1.2 to 3.5

1.8 to 2.7

.01
.17
.009

Abbreviations: HPV, human papillomavirus; HR, hazard ratio; FIGO, International Federation of Gynecology and Obstetrics; LN, lymph node.

Burger et al12 also identified HPV 18 positivity as a significant

poor prognostic factor in FIGO stage I and II patients (n 171)
undergoing RH-PLND, whereas in the analysis of the whole series of
patients with stage I to IV disease (n 291), only stage was significant.
Schwartz et al14 noted that HPV 18 related tumors were significantly
related to poor outcome for patients with FIGO stage IB and IIA but
not prognostic for stage IB to IV. Viladiu et al16 also identified clinical
stage as the only independent prognostic factor, and HPV 16 E7
antibody predicted mortality only limited to stage I and II.
Lombard et al13 confirmed the adverse prognostic significance of
HPV 18 positivity in 399 patients with stage I to IV disease; however,
the variables included in the multivariate analysis were limited to those
obtainable without RH-PLND. Conversely, Plich et al17 identified
HPV 16 positivity, not HPV 18 positivity, as a poor prognostic factor
in 204 patients undergoing primary RH-PLND.
In this series, HPV 58, 33, or 52 was associated with older age and
squamous carcinoma. The 5-year RFS rate of patients positive for
HPV 58, 33, or 52 (n 165) was marginally higher than the rate for
patients negative for HPV 58, 33, or 52 (P .07) by univariate analysis.
In this study, HPV 58, 33, or 52 positivity was detected in 16.5% and
multiple-type HPV was detected in 9.6% of early-stage primarily
surgically treated patients compared with 44.5% and 26.5% of patients, respectively, receiving RT/CCRT in our previous study.24
Regarding the prognostic impact of multiple infections, few data
are available in the literature. Moreover, multiple infections cannot be

the original models for either age cutoff at 50 or 45 years (Appendix

Table A3), except that tumor size was selected as a significant predictor
for RFS (Appendix Table A4). However, tumor size was highly correlated with FIGO stage, depth of stromal invasion, parametrial extension, lymph node metastasis, and grade of differentiation (data not
shown). Adding variable tumor size into the proposed model or using
the exact HR to formulate alternative models did not fit the model any
better in the process of model building.
DISCUSSION

The independent prognostic significance of HPV 18 positivity in

early-stage cervical cancer is confirmed in this sufficiently sized study.
HPV 18 positivity was significantly more common in adenoadenosquamous carcinoma and lymphatic permeation but was unrelated to age, tumor size, FIGO stage, depth of cervical stromal
invasion, lymph node metastasis, and parametrial extension.
In this series, FIGO stage, lymph node metastasis, depth of cervical stromal invasion more than one third, grade 2 to 3 differentiation,
and HPV 18 positivity were significant predictors of relapse, and FIGO
stage II, depth of cervical stromal invasion more than one third,
parametrial extension, HPV 18 positivity, and age older than 45 years
were significant predictors for death by multivariate analysis. The
prognostic models including HPV 18 positivity are powerful in predicting death (P .0001) and relapse (P .0001).

Table 4. HRs of Death and Relapse According to Prognostic Score

Death

Risk Group

Prognostic
Score

No. of
Patients

5-Year
OS (%)

HR

95% CI

Low
Intermediate
High

0-2
3
4-7

565
319
183

97.4
87.6
77.9

1
5.0
7.8

Reference
3.1 to 8.1
4.8 to 12.7

Relapse

5-Year
RFS (%)

HR

95% CI

.0001
.0001

95.9
86.7
72.7

1
3.4
6.5

Reference
2.1 to 5.2
4.1 to 10.1

.0001
.0001

Abbreviations: HR, hazard ratio; OS, overall survival; RFS, recurrence-free survival.

3632

JOURNAL OF CLINICAL ONCOLOGY

HPV Genotype and Cervical Cancer Prognosis

1.0
0.9

Overall Survival

0.8
0.7
0.6
0.5
0.4

Low risk
Intermediate risk
High risk

0.3
0.2
0.1
0

P < .0001

10 11 12 13 14

Time (years)

B
Recurrence-Free Survival

1.0
0.9
0.8
0.7
0.6

was weakened despite bootstrap resampling to reduce bias. In addition, we gave equal weight to each predictor that had a different HR
(1.7 to 3.2) to facilitate clinical use. We attempted to use a sample of
patients (n 274) treated between 2001 and 2003 for external validation; however, the number of events was small (20 recurrences and 14
deaths), and the median follow-up time was relatively short, which
made it invalid.
Among the seven selected risk factors, only grade, FIGO stage,
HPV genotype, and age can be accurately assessed at preoperative
period, whereas clinical occult lymph node and/or parametrial metastasis or depth of cervical stromal invasion cannot be reliably determined by imaging technology.34,35 In this series, for patients who have
three risk factors (n 80), additional HPV genotyping can help to
exclude HPV 18 positive patients from primary radical surgery because four risk factors would predict a high probability of postoperative RT/CCRT. We should avoid using both modalities in early-stage
cervical cancer. Moreover, for patients who have three risk factors
other than HPV genotype after primary surgery (n 318), adjuvant
therapy can be offered to patients with HPV 18 positive tumors.
In conclusion, HPV genotyping is worthwhile to perform because of its independent prognostic value in early-stage cervical cancer, and the predicting models for death and relapse could be useful for
counseling the individual patient and stratifying study patients in
future clinical trials.

0.5
0.4

Low risk
Intermediate risk
High risk

0.3

AUTHORS DISCLOSURES OF POTENTIAL CONFLICTS

OF INTEREST

0.2
0.1
0

The author(s) indicated no potential conflicts of interest.

P < .0001

Time (years)
Fig 1. (A) Kaplan-Meier overall survival curves according to risk groups in
early-stage cervical cancer patients undergoing primary surgery (n 1,067). (B)
Kaplan-Meier recurrence-free survival curves according to risk groups in earlystage cervical cancer patients undergoing primary surgery (n 1,067).

identified using direct sequencing; therefore, those using only direct

sequencing tend to severely underestimate the frequency of multipletype HPV.24,26,31,32 Bachtiary et al33 reported a study of 106 cervical
cancer patients receiving RT, in which multiple-type HPV was an
independent poor prognostic factor. In the current study, there was no
correlation of multiple-type HPV with RFS or OS in patients receiving
primary surgery. The prognostic significance of multiple HPV types in
RT/CCRT patients will be answered in our next report.
Limitations of our study include that we did not test the predicting models using an external validation data set. Statistical inference
REFERENCES
1. Waggoner SE: Cervical cancer. Lancet 361:
2217-2225, 2003
2. Benedet JL, Bender H, Jones H 3rd, et al:
FIGO staging classifications and clinical practice guidelines in the management of gynecologic cancers: FIGO Committee on Gynecologic
Oncology. Int J Gynaecol Obstet 70:209-262,
2000
www.jco.org

AUTHOR CONTRIBUTIONS

10 11 12 13 14

Conception and design: Chyong-Huey Lai, Chee-Jen Chang, Huei-Jean

Huang
Provision of study materials or patients: Chyong-Huey Lai, Huei-Jean
Huang, Swei Hsueh, Angel Chao, Cheng-Tao Lin, Ji-Hong Hong,
Hung-Hsueh Chou, Tzu-I Wu, Kuan-Gen Huang, Chun-Chieh Wang,
Ting-Chang Chang
Collection and assembly of data: Chyong-Huey Lai, Huei-Jean Huang,
Swei Hsueh, Angel Chao, Jung-Erh Yang, Cheng-Tao Lin, Shang-Lang
Huang, Ji-Hong Hong, Hung-Hsueh Chou, Tzu-I Wu, Kuan-Gen
Huang, Chun-Chieh Wang, Ting-Chang Chang
Data analysis and interpretation: Chyong-Huey Lai, Chee-Jen Chang,
Huei-Jean Huang
Manuscript writing: Chyong-Huey Lai, Chee-Jen Chang, Huei-Jean
Huang
Final approval of manuscript: Chyong-Huey Lai, Chee-Jen Chang,
Huei-Jean Huang, Swei Hsueh, Angel Chao, Jung-Erh Yang, Cheng-Tao
Lin, Shang-Lang Huang, Ji-Hong Hong, Hung-Hsueh Chou, Tzu-I Wu,
Kuan-Gen Huang, Chun-Chieh Wang, Ting-Chang Chang

3. Landoni F, Maneo A, Colombo A, et al: Randomized study of radical surgery versus radiotherapy for stage IB-IIA cervical cancer. Lancet 350:535540, 1997
4. Delgado G, Bundy B, Zaino R, et al: Prospective surgical-pathological study of disease-free interval in patients with stage IB squamous carcinoma of
the cervix: A Gynecologic Oncology Group study.
Gynecol Oncol 38:352-357, 1990
5. Lai CH, Hong JH, Hsueh S, et al: Preoperative
prognostic variables and impacts of postoperative

adjuvant therapy in stage IB or II cervical carcinoma

with or without pelvic node metastases: An analysis
of 891 patients. Cancer 85:1537-1546, 1999
6. zur Hausen H: Papillomaviruses causing
cancer: Evasion from host-cell control in early
events in carcinogenesis. J Natl Cancer Inst 92:
690-698, 2000
7. Munoz N, Bosch FX, de Sanjose S, et al:
Epidemiologic classification of human papillomavirus types associated with cervical cancer. N Engl
J Med 348:518-527, 2003
3633

Lai et al

8. Walboomers JM, Jacobs MV, Manos MM, et

al: Human papillomavirus is a necessary cause of
invasive cervical cancer worldwide. J Pathol 189:1219, 1999
9. van Muyden RC, ter Harmsel BW, Smedts
FM, et al: Detection and typing of human papillomavirus in cervical carcinomas in Russian women: A
prognostic study. Cancer 85:2011-2016, 1999
10. Rose BR, Thompson CH, Simpson JM, et al:
Human papillomavirus deoxyribonucleic acid as a
prognostic indicator in early stage cervical cancer: A
possible role for type 18. Am J Obstet Gynecol
173:1461-1468, 1995
11. Nakagawa S, Yoshikawa H, Onda T, et al:
Type of human papillomavirus is related to clinical
features of cervical carcinoma. Cancer 78:19351941, 1996
12. Burger RA, Monk BJ, Kurosaki T, et al: Human
papillomavirus type 18: Association with poor prognosis in early stage cervical cancer. J Natl Cancer
Inst 88:1361-1368, 1996
13. Lombard I, Vincent-Salomon A, Validire P, et
al: Human papillomavirus genotype as a major determinant of the course of cervical cancer. J Clin
Oncol 16:2613-2619, 1998
14. Schwartz SM, Daling JR, Shera KA, et al:
Human papillomavirus and prognosis of invasive
cervical cancer: A population-based study. J Clin
Oncol 19:1906-1915, 2001
15. Im SS, Wilczynski SP, Burger RA, et al: Early
stage cervical cancers containing human papillomavirus type 18 DNA have more nodal metastasis and
deeper stromal invasion. Clin Cancer Res 9:41454150, 2003
16. Viladiu P, Bosch FX, Castellsague X, et al:
Human papillomavirus DNA and antibodies to human papillomaviruses 16 E2, L2, and E7 peptides as
predictors of survival in patients with squamous cell
cervical cancer. J Clin Oncol 15:610-619, 1997
17. Pilch H, Gunzel S, Schaffer U, et al: The
presence of HPV DNA in cervical cancer: Correlation
with clinicopathologic parameters and prognostic

significance10 years experience at the Department of Obstetrics and Gynecology of the Mainz
University. Int J Gynecol Cancer 11:39-48, 2001
18. Graflund M, Sorbe B, Sigurdardottir S, et al:
HPV-DNA, vascular space invasion, and their impact
on the clinical outcome in early-stage cervical carcinomas. Int J Gynecol Cancer 14:896-902, 2004
19. Dahlgren L, Erlandsson F, Lindquist D, et al:
Differences in human papillomavirus type may influence clinical outcome in early stage cervical cancer.
Anticancer Res 26:829-832, 2006
20. Huang LW, Chao SL, Hwang JL: Human
papillomavirus-31-related types predict better survival in cervical carcinoma. Cancer 100:327-334,
2004
21. Lai HC, Sun CA, Yu MH, et al: Favorable
clinical outcome of cervical cancers infected with
human papilloma virus type 58 and related types. Int
J Cancer 84:553-557, 1999
22. van Bommel PF, van den Brule AJ, Helmerhorst TJ, et al: HPV DNA presence and HPV genotypes as prognostic factors in low-stage squamous
cell cervical cancer. Gynecol Oncol 48:333-337,
1993
23. Fule T, Csapo Z, Mathe M, et al: Prognostic
significance of high-risk HPV status in advanced
cervical cancers and pelvic lymph nodes. Gynecol
Oncol 100:570-578, 2006
24. Lai CH, Huang HJ, Hsueh S, et al: Human
papillomavirus genotype in cervical cancer: A
population-based study. Int J Cancer 120:19992006, 2007
25. Huang HJ, Huang SL, Lin CY, et al: Human
papillomavirus genotyping by a polymerase chain
reaction-based genechip method in cervical carcinoma treated with neoadjuvant chemotherapy plus
radical surgery. Int J Gynecol Cancer 14:639-649,
2004
26. Huang SL, Chao A, Hsueh S, et al: Comparison between the Hybrid Capture II Test and a
SPF1/GP6 PCR-based assay for detection of hu-

man papillomavirus DNA in cervical swab samples.

J Clin Microbiol 44:1733-1739, 2006
27. Lai CH, Hsueh S, Chang TC, et al: Implications
of a failed trial of postoperative adjuvant therapy
after radical hysterectomy for stage IB-IIA cervical
carcinoma with pelvic node metastases. Changgeng
Yi Xue Za Zhi 21:291-299, 1998
28. Hong JH, Tsai CS, Lai CH, et al: Postoperative
low-pelvic irradiation for stage I-IIA cervical cancer
patients with risk factors other than pelvic lymph
node metastasis. Int J Radiat Oncol Biol Phys 53:
1284-1290, 2002
29. Lai CH: Management of recurrent cervical
cancer. Chang Gung Med J 72:711-719, 2004
30. Steyerberg EW, Harrell FE Jr, Borsboom GJ,
et al: Internal validation of predictive models: Efficiency of some procedures for logistic regression
analysis. J Clin Epidemiol 54:774-781, 2001
31. Molijn A, Kleter B, Quint W, et al: Molecular
diagnosis of human papillomavirus (HPV) infection.
J Clin Virol 32S:S43-S51, 2005 (suppl 1)
32. Lo KW, Wong YF, Chan MK, et al: Prevalence
of human papillomavirus in cervical cancer: A multicenter study in China. Int J Cancer 100:327-331,
2002
33. Bachtiary B, Obermair A, Dreier B, et al:
Impact of multiple HPV infection on response to
treatment and survival in patients receiving radical
radiotherapy for cervical cancer. Int J Cancer 102:
237-243, 2002
34. Mitchell DG, Snyder B, Coakley F, et al: Early
invasive cervical cancer: Tumor delineation by magnetic resonance imaging, computed tomography,
and clinical examination, verified by pathologic results, in the ACRIN 6651/GOG 183 Intergroup study.
J Clin Oncol 24:5687-5694, 2006
35. Chou HH, Chang TC, Yen TC, et al: Low value
of 2-fluoro-2-deoxy-d-glucose positron emission tomography in primary staging of early-stage cervical
cancer prior to radical hysterectomy. J Clin Oncol
24:123-128, 2006

Appendix
The Appendix is included in the full-text version of this article, available online at www.jco.org. It is not included in the PDF version
(via Adobe Reader).

3634

JOURNAL OF CLINICAL ONCOLOGY