You are on page 1of 13

Journal of International Medical Research http://imr.sagepub.

com/

An Efficacy Analysis for Nasopharyngeal Carcinoma Screening of Different Screening Intervals


DP Rao, J Gu, XH Meng, Q Zhang, MQ Fu, Y Liu, F Chen, SM Cao, MH Hong and Q Liu Journal of International Medical Research 2012 40: 525 DOI: 10.1177/147323001204000214 The online version of this article can be found at: http://imr.sagepub.com/content/40/2/525

Published by:
http://www.sagepublications.com

Additional services and information for Journal of International Medical Research can be found at: Email Alerts: http://imr.sagepub.com/cgi/alerts Subscriptions: http://imr.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav

>> Version of Record - Apr 1, 2012 What is This?

Downloaded from imr.sagepub.com by guest on October 5, 2013

The Journal of International Medical Research 2012; 40: 525 536

An Efficacy Analysis for Nasopharyngeal Carcinoma Screening of Different Screening Intervals


DP RAO1,2,a, J GU1,a, XH MENG3, Q ZHANG3, MQ FU4, Y LIU2, F CHEN2, SM CAO2, MH HONG2,5 AND Q LIU2
1

Department of Medical Statistics and Epidemiology, and 3Department of Health Service Management, School of Public Health, Sun Yat-sen University, Guangzhou, China; 2 Department of Epidemiology, and 5Department of Clinical Trials, Cancer Centre, Sun Yat-sen University, Guangzhou, China; 4Department of Microbiology, School of Public Health and Tropical Medicine, Southern Medical University, Guangzhou, China rate with strategy A1 (annual screening) were 23.6% and 83.9%, respectively. Compared with strategy A1, strategy B1 (annual screening for seropositive subjects; biennial screening for seronegative subjects) had a similar 5-year mortality rate (24.0%) and a slightly smaller NPC pick-up rate (81.7%), but led to a 39.3% reduction in total screenings. Compared with all other strategies excluding strategy A1, strategy B1 achieved the lowest 5-year mortality rate and the largest NPC pick-up rate. CONCLUSIONS: Strategy B1 had the highest efficacy for NPC screening.

OBJECTIVE: To evaluate the impact of different screening intervals on screening for nasopharyngeal carcinoma (NPC). METHODS: A Markov model was constructed, based on the natural history of NPC. The 5-year mortality rate of NPC was the major measurement to evaluate the efficacies of 16 screening strategies. Parameters for the model were derived from published literature. RESULTS: Screening reduced the 5-year mortality rate for NPC by 20.4 43.3%, compared with the equivalent rate without screening. The 5year mortality rate and the NPC pick-up

KEY WORDS: NASOPHARYNGEAL CARCINOMA; SCREENING; MARKOV MODEL; SIMULATION; EFFICACY; EPSTEINBARR VIRUS (EBV) SCREENING INTERVAL; EBV-SEROPOSITIVE SUBJECTS; EBV-SERONEGATIVE SUBJECTS

Introduction
Nasopharyngeal carcinoma (NPC) occurs rarely in most parts of the world but its prevalence is high in southeast Asia and China.1 5 The reported incidence of NPC in the Chinese population is 10 40/100 000 per year,1,6 8 whereas it is < 6/100 000 per year in North AfricanArabic populations9
a

DP Rao and J Gu contributed equally to this work.

and < 0.3/100 000 per year in Japan.10 Among the Chinese provinces, Guangdong ranks first in terms of the incidence of NPC, as reported by the Chinese Cancer Registry System.1,4,5,11 EpsteinBarr virus (EBV) infection is strongly associated with the occurrence of NPC,5,12 20 and the immunoglobulin (Ig)A antibody against the EBV viral capsid antigen (EBV VCA/IgA) is a widely used antibody

525

DP Rao, J Gu, XH Meng et al. Efficacy analysis for NPC screening


marker for NPC screening.21 25 Studies have shown that screening for NPC is an effective measure for secondary prevention.6,25 28 A clinical and serological follow-up study of NPC conducted between 1986 and 2002 in Zhongshan City, Guangdong Province, China, reported that two-thirds of NPC cases could be detected at an early disease stage (i.e. 1997 American Joint Committee of Cancer/International Union Against Cancer tumournodemetastasis classification system stages I II),29,30 using an EBV VCA/IgA test.25 The proportion of early-stage NPC cases detected using an EBV VCA/IgA test was significantly higher than the proportion of symptomatic cases detected by the same method (i.e. 67.6% and 19.6%, respectively).25 Markov processes, as a modelling technique, have been widely applied in evaluations of cancer screening.31 33 Only two studies, however, have used simulated models to evaluate screening strategies for NPC. In 1999, Chen et al.27 constructed a Markov model of the natural history of NPC, using published screening data from the general population of Guangxi Province in China, to predict deaths caused by the disease. The results showed that the mortality rate in patients with NPC who were screened with the EBV VCA/IgA test was sharply reduced (by 30%) compared with the mortality rate in patients who did not receive screening.27 In 2010, Choi et al.28 used Markov chain models and reported that the 5-year survival rate for familial NPC with screening varied from 77.7% to 80.1%, which was higher than that observed without screening (67.9%). Though both studies illustrated that the survival rate of NPC patients would be improved by screening,27,28 these findings were based on the assumption that, in the natural history model of NPC, regression from a state of EBV-seropositivity to a state of health was not possible.27,28 There is growing evidence, however, that EBV seropositivity can regress to a state of health.25,34 One study reported that 10.9% of EBV antibody-positive individuals became EBV antibody-negative, and that 5.4% of EBV-seronegative individuals became EBVseropositive, within 4 years.34 The screening interval for NPC is a controversial issue that has been inadequately evaluated. Chen et al.27 considered four screening strategies: 3-year IgA/VCA test plus annual indirect mirror examination; 3-year IgA/VCA test plus 3-year indirect mirror examination; 6-year IgA/VCA test plus annual indirect mirror examination; and 6-year IgA/VCA test plus 3-year indirect mirror examination. They reported no significant differences between the 3- and 6year serum screenings.27 Choi et al.28 evaluated four strategies: annual screening; biennial screening; triennial screening; and annual screening for EBV-seropositive subjects plus triennial screening for EBV-seronegative subjects. They found that triennial screening for EBV-seronegative subjects and annual screening for EBV-seropositive subjects had the highest efficacy for NPC screening.28 Because of the lack of consensus on the appropriate screening interval for NPC, it is necessary to take other screening strategies into account in order to evaluate fully the efficacy of different approaches. Furthermore, validation analyses for the natural history model of NPC and sensitivity analyses were not performed in the previous studies.27,28 A Markov model was constructed for the present study that allowed for regression from a state of EBV-seropositivity to a state of health. A validation analysis for the natural history model of NPC and a sensitivity analysis were also performed. In addition, the screening strategies for NPC were altered by using different time intervals and a total of 16 strategies were evaluated.

526

DP Rao, J Gu, XH Meng et al. Efficacy analysis for NPC screening

Materials and methods


SIMULATION MODEL
To compare the efficacies of different screening strategies for NPC, a Markov simulation model (based on the natural history of NPC) was constructed using SAS software, version 9.2 (SAS Institute, Cary, NC, USA) for Windows. The Markov simulation model consisted of two parts: the first part involved simulation of the natural history of NPC among a high-risk population; the second part simulated NPC screening. The simulation followed a cohort of 100 000 individuals as they aged from 30 to 59 years. Each individual would move through the simulation until they ended in the absorbing state. The Markov model is shown in Fig. 1. Parameters for simulation (such as EBVseropositive rate, sensitivity and specificity of the EBV VCA/IgA test, 5-year survival rate for NPC and NPC incidence), were sourced from Sihui City in Guangdong Province, China. Annual transition probabilities for the natural history model were obtained from published literature.27,34 The 5-year mortality rate, the total number of screens and the NPC pick-up rate were estimated for

each strategy in the present study. The NPC pick-up rate was calculated to be the number of detected NPC incidences divided by the total number of NPC incidences. The 5-year mortality rates were based on the 5-year survival rates of screen-detected and symptomatic NPC, as shown in Formula 1. The 5-year survival rates of screen-detected and symptomatic NPC used in the present study were 79.9% and 58.4%, respectively (Formula 1).35

NATURAL HISTORY MODEL


The natural history model of NPC was based on a previous study,27 but also assumed that the state of EBV-positivity could regress to the state of health during the course of a year (Fig. 1).25,34 The natural history states of NPC were: health; EBV-seropositive; preclinical but screen-detectable phase (PCDP); and clinical NPC. The state of health indicated that the subject did not have NPC and that EBV antibodies could not be detected in serum. EBV-seropositive indicated that EBV antibodies could be found on serological testing, but that PCDP had not developed. PCDP included individuals in whom NPC was

5-year mortality rate = 1

No. of screen-detected No. of symptomatic 79.9% + 58.4% NPC NPC No. of total NPC

FORMULA 1: Calculation of 5-year mortality rate from nasopharyngeal carcinoma (NPC)

Health

EBV-seropositive

PCDP

Clinical NPC

Detection from screening FIGURE 1: Markov model incorporating simulation of the natural history of nasopharyngeal carcinoma (NPC) and screening for NPC. EBV, EpsteinBarr virus; PCDP, preclinical but screen-detectable phase

527

DP Rao, J Gu, XH Meng et al. Efficacy analysis for NPC screening


detected, but who were asymptomatic at screening. Clinical NPC described individuals with symptoms of NPC. Subjects in the state of health could remain in the same state or progress to the state of being EBVseropositive. Subjects in the state of EBVseropositivity could remain in the same state, progress to the state of PCDP or regress to the state of health. Subjects in the state of PCDP could remain in the same state or progress to clinical NPC. Clinical NPC was a surrogate endpoint, similar to that used by Chen et al.27 remainder (92.815%) were in the state of health.

SCREENING STRATEGIES
According to the NPC screening process currently used in Sihui City, EBV status is examined by serological testing against EBV VCA/IgA; any subject identified as being EBV-seropositive then has a nasopharyngeal fibrescope (NPF) examination. If the result of the NPF examination is positive, the subject undergoes a biopsy to diagnose NPC. The present study compared 16 NPC screening strategies (Table 2): the screening interval differed in each strategy but other parameters were unchanged. Assumptions were that all subjects complied with the programme, and that the sensitivity and specificity rates of the NPF examination were 98% and 100%, respectively. The sensitivity and specificity rates of the EBV VCA/IgA test were 95% and 90%, respectively.37

PROBABILITIES OF TRANSITION
The probabilities of transition among the states of the natural history model of NPC were obtained in the present study by two different measures. Using the method of Miller and Homan,36 the annual transition probabilities from health to EBV-seropositive and from EBVseropositive to health were determined from transition rates obtained from the literature.34 Other annual transition probabilities were obtained from the study by Chen et al.27 The probabilities are shown in Table 1. The simulation began with a cohort of 100 000 30year-old subjects who were followed for 30 years (i.e. until age 59). Assumptions were made that, in the initial population, 7.160% of the subjects were in the EBV-seropositive state,24 0.025% were in the PCDP state11 and the

MODEL VALIDATION
Validation of the model involved calibration by comparing the predicted age-specific incidences of NPC with the observed NPC incidences in Sihui City in 2005.38 Prediction of the age-specific incidences of NPC was based on the natural history model of NPC. Pearsons 2-test was performed to

TABLE 1: Annual transition probabilities for the natural history model of nasopharyngeal carcinoma (NPC) Transition Health to EBV-seropositive Health to PCDP Health to clinical NPC EBV-seropositive to health EBV-seropositive to PCDP EBV-seropositive to clinical NPC PCDP to clinical NPC Annual transition probability 0.013782 0.00000095 0.000000106 0.028440 0.002402273 0.000412757 0.27805 Reference Zeng et al.34 Chen et al.27 Chen et al.27 Zeng et al.34 Chen et al.27 Chen et al.27 Chen et al.27

EBV, EpsteinBarr virus; PCDP, preclinical but screen-detectable phase.

528

DP Rao, J Gu, XH Meng et al. Efficacy analysis for NPC screening

TABLE 2: The 16 strategies that were used to screen for nasopharyngeal carcinoma in EpsteinBarr virus (EBV)-seropositive and EBV-seronegative subjects Strategy A1 A2 A3 A4 A5 A6 B1 B2 B3 B4 B5 C1 C2 C3 C4 C5 Event(s) Annual screening Biennial screening Triennial screening 4-year screening 5-year screening 6-year screening Annual screening for seropositive subjects; biennial screening for seronegative subjects Annual screening for seropositive subjects; triennial screening for seronegative subjects Annual screening for seropositive subjects; 4-year screening for seronegative subjects Annual screening for seropositive subjects; 5-year screening for seronegative subjects Annual screening for seropositive subjects; 6-year screening for seronegative subjects Annual screening for seronegative subjects; biennial screening for seropositive subjects Annual screening for seronegative subjects; triennial screening for seropositive subjects Annual screening for seronegative subjects; 4-year screening for seropositive subjects Annual screening for seronegative subjects; 5-year screening for seropositive subjects Annual screening for seronegative subjects; 6-year screening for seropositive subjects

investigate matching between the predicted and observed incidences of NPC in Sihui City: the Sihui Cancer Registry System was established in 1978 and records all cancer patients from each regional hospital in Sihui and from the local clinical oncology departments of the citys hospitals.35 Since 1987, Sihui City has been one of the epidemiological fields where the Chinese National Programmes for Science and Technology Development has launched initiatives to improve early detection of NPC by serological screening tests.

SENSITIVITY ANALYSES
One-way sensitivity analyses were performed to assess the sensitivity of the mortality rate to changes in screening test sensitivity, screening interval and compliance of the subject.

Results
PREDICTED OUTCOMES
Predicted outcomes are shown in Table 3. In the case of no screening, the model predicted

1511 cases of NPC in the simulated population of 100 000 adults, which corresponded to a 5-year mortality rate of 41.6%. The 5-year mortality rate for NPC increased, whereas the NPC pick-up rate and the total screens decreased with increasing screening interval (Table 3). Compared with no screening, all 16 strategies improved life expectancy, with a decrement of > 20% in the 5-year mortality rate (Table 3). Strategy A1 (annual screening) achieved the lowest 5year mortality rate (23.6%), the highest NPC pick-up rate (83.9%) and the largest number of total screens (3 081 193). Compared with strategy A1, the 5-year mortality rate with strategy B1 (annual screening for seropositive subjects; biennial screening for seronegative subjects) slightly increased from 23.6% to 24.0% and the NPC pick-up rate slightly decreased from 83.6% to 81.7%; however, the total screens dramatically decreased by 39.3%, from 3 081 193 to 1 870 554. The 5year mortality rate with strategy B1 decreased sharply by 42.3%, from 41.6% to

529

DP Rao, J Gu, XH Meng et al. Efficacy analysis for NPC screening

TABLE 3: Predicted outcomes of the 16 screening strategies (defined in Table 2) in detecting nasopharyngeal carcinoma (NPC) Strategy A1 A2 A3 A4 A5 A6 B1 B2 B3 B4 B5 C1 C2 C3 C4 C5 No screening Detected Missed cases of NPC cases of NPC 1267 1090 930 742 720 653 1235 1178 1139 1091 1066 1042 904 770 678 594 244 421 581 771 791 858 276 333 372 420 445 469 607 741 833 917 Total screens 3 081 193 1 590 698 1 093 749 795 950 696 104 596 673 1 870 554 1 417 311 1 174 386 1 014 740 903 954 2 593 772 2 319 610 2 121 816 1 967 385 1 840 692 NPC pick-up 5-year mortality rate, % rate, % 83.9 72.1 61.5 49.1 47.7 43.2 81.7 78.0 75.4 72.2 70.5 69.0 59.8 51.0 44.9 39.3 23.6 26.1 28.4 31.1 31.4 32.3 24.0 24.8 25.4 26.1 26.4 26.8 28.7 30.6 32.0 33.1 41.6

24.0%, compared with no screening. Furthermore, compared with strategy B1, all strategies except strategy A1 resulted in increases in the 5-year mortality rate and decreases in the NPC pick-up rate (Table 3).

MODEL VALIDATION
In the model, the number of predicted NPC incidences rose from 20.7 to a peak of 69.0 per 100 000 per year as the subjects ages increased from 30 to 59 years (Fig. 2). In 2005, the number of NPC incidences in Sihui City increased with age, and a peak incidence of 75 per 100 000 per year was observed at age 59.38 Differences between the predicted NPC incidence and the observed NPC incidence in Sihui City were not statistically significant.

SENSITIVITY ANALYSES
One-way sensitivity analysis was performed to assess the sensitivity of the mortality rate to changes in screening test sensitivity, screening interval and compliance of the subject.

Baseline parameters and assumptions for the model are outlined in Table 4. When one of the baseline parameters changed, the other parameters were maintained at the baseline level. In this analysis, the model was sensitive to variations in the sensitivity of the EBV VCA/IgA test, screening interval and compliance of the subject. Decreased sensitivity of the screening test was related to an increase in the 5-year mortality rate (Fig. 3). Variation in the sensitivity of the screening test from 70% to 99% resulted in an estimated increment of the 5-year mortality rate of 9.7%: the actual 5-year mortality rate went from 24.7% for the highest sensitivity to 27.1% for the lowest sensitivity. The 5-year mortality rate also increased with an increase in the screening interval (Fig. 4). Increasing the screening interval from every 1 year to every 6 years, increased the 5-year mortality rate from 25.1% to 34.1%. A decrease in NPC screening compliance resulted in an increase in the 5-

530

DP Rao, J Gu, XH Meng et al. Efficacy analysis for NPC screening

80 Observed in Sihui in 200538 Predicted in Markov model NPC incidence per 100 000 per year

60

40

20

30 34

35 39

40 44 45 49 Age (years)

50 54

55 59

FIGURE 2: Incidence of nasopharyngeal carcinoma (NPC) in Sihui CIty, China, in 200538 compared with the incidence the predicted in the Markov model, according to age of the subjects. Differences between predicted and observed incidences were not statistically significant (P 0.05; Pearsons 2-test) year mortality rate, which ranged from 23.6% with 100% compliance to 28.0% for 50% compliance (Fig. 5). strategies has not been adequately evaluated. An improved Markov simulation model was developed in the present study, in order to evaluate screening strategies for NPC using different time intervals. All the screening strategies used for NPC in the present study reduced the mortality rate and increased the pick-up rate for NPC. With screening, the 5-year mortality rate for NPC was reduced by 20.4 43.3% compared with

Discussion
Previous studies have shown that screening for NPC can significantly reduce the mortality rate associated with this disease,27,28,35 but the impact of screening interval on the efficacy of different screening

TABLE 4: Baseline parameters and assumptions for the screening model for nasopharyngeal carcinoma (one-way sensitivity analysis) Screening parameter Sensitivity of the EBV VCA/IgA test Screening interval Screening compliance Baseline parameter value 95%37 1 year 80% Sensitivity analysis range 70 99% 1 year to every 6 years 50 100%

EBV, EpsteinBarr virus; VCA, viral capsid antigen; IgA, immunoglobulin A.

531

DP Rao, J Gu, XH Meng et al. Efficacy analysis for NPC screening

27.5

27.0

5-year mortality rate (%)

26.5

26.0

25.5

25.0

24.5 70 75 80 85 90 Sensitivity of NPC screening (%) 95 100

FIGURE 3: Sensitivity analysis for the 5-year mortality rate from nasopharyngeal carcinoma (NPC), with various sensitivities applied to the screening test in the Markov model for the natural history of NPC the mortality rate without screening. The NPC pick-up rate with screening varied from 39.3% to 83.9%. An improved model of the natural history of NPC was constructed successfully. The model in the present study was more complete than those used previously.27,28 The present model assumed that the state of EBVseropositivity could regress to the state of health during the course of a year. As validation for the model, the number of predicted NPC incidences in the model was shown to be consistent with the number of incidences observed in Sihui City in 2005.38 The present study suggested that strategy B1 had the highest efficacy for NPC screening. Using strategy A1 for NPC screening could achieve the lowest 5-year mortality rate and the highest NPC pick-up rate of the 16 strategies evaluated, but resulted in the largest number of screens over a 30-year period. Compared with strategy A1, strategy B1 had a similar 5-year mortality rate (24.0%) and a similar NPC pick-up rate (81.7%). Notably, however, the total number of screens with strategy B1 (1 870 554) was decreased by 39.3% from that for strategy A1 (3 081 193). Additionally, all of the remaining strategies (excluding A1) led to higher 5-year mortality rates and lower NPC pick-up rates. This suggests that strategy B1 could achieve the highest efficacy for NPC screening. This finding was not consistent with that of a study undertaken in Hong Kong28 in which four strategies (equivalent to A1, A2, A3 and B2 in the present study) were considered. In the Hong Kong study, strategy B1 was not evaluated, and strategy B2 (annual screening for seropositive subjects; triennial screening for seronegative subjects) had a similar 5-year mortality rate to that of

532

DP Rao, J Gu, XH Meng et al. Efficacy analysis for NPC screening

36.0

34.0

5-year mortality rate (%)

32.0

30.0

28.0

26.0

24.0 1 2 3 4 Screening interval (years) 5 6

FIGURE 4: Sensitivity analysis for the 5-year mortality rate from nasopharyngeal carcinoma (NPC) at various screening intervals in the Markov model for the natural history of NPC

strategy A1.28 In the present study, the efficacies of strategies A3 (triennial screening) and A6 (6-year screening) were different and did not correspond to those of a previous study published in 1999.27 The present study found that the impact of screening intervals for EBV-seropositive and EBV-seronegative subjects was different and had a considerable effect on the mortality rate from NPC. When the interval for EBV-seropositive subjects was fixed at 1 year and the interval for EBV-seronegative subjects varied from 2 to 6 years (i.e. strategies B1 to B5), the 5-year mortality rate for NPC increased by 10.0% (from 24.0% to 26.4%), the NPC pick-up rate decreased by 13.7% (from 81.7% to 70.5%) and the total screens decreased by 51.7% (from 1 870 554

to 903 954). When the interval for EBVseronegative subjects was fixed at 1 year and the interval for EBV-seropositive subjects changed from 2 to 6 years (strategies C1 to C5), the 5-year mortality rate for NPC increased by 23.5% (from 26.8% to 33.1%), the NPC pick-up rate decreased by 43.0% (from 69.0% to 39.3%), and the total screens decreased by 29.0% (from 2 593 772 to 1 840 692). These changes showed that, for EBV-seropositive subjects, the impact of screening interval on the 5-year mortality rates and the NPC pick-up rates was stronger than for the EBV-seronegative subjects, and that the impact of intervals for EBVseronegative subjects on the total screens was stronger than for EBV-seropositive subjects. Consequently, to reduce the total

533

DP Rao, J Gu, XH Meng et al. Efficacy analysis for NPC screening

29.0

28.0 5-year mortality rate (%)

27.0

26.0

25.0

24.0

23.0 50 60 70 80 Compliance (%) 90 100

FIGURE 5: Sensitivity analysis for the 5-year mortality rate from nasopharyngeal carcinoma (NPC) at various levels of screening compliance in the Markov model for the natural history of NPC screens for NPC, the interval for EBVseronegative subjects should be long and, to improve the mortality rate and pick-up rate of NPC, the interval for EBV-seropositive subjects should be short. The present study suggested that sensitivity of the screening test, screening interval and screening compliance could affect the mortality rate for NPC. In fact, these factors can change the number of patients with NPC who are detected by screening. When the sensitivity of the screening test or the screening compliance was improved, the number of patients detected should increase. A shorter interval should lead to a larger number of patients with NPC being detected by screening and should improve the 5-year survival rate for this disease. With screening, more cases of NPC can be detected at an early disease stage compared with the number detected without screening.25 28,39 The effects of treatment are better for early-stage than for advanced NPC;6,40 in addition, the survival rate for patients with NPC detected by screening is better than that for patients who present with symptomatic NPC.26 28,35,41 As a result, to yield more patients with early-stage NPC, the sensitivity of the screening test and screening compliance should be at a high level and, if resources are sufficient, the screening interval should be as short as possible. There were some limitations to the present study. First, the predicted outcomes of the 16 screening strategies for NPC were simulated assuming that subject compliance in the screening programme was 100%. Compliance of subjects can impact the efficacy of screening for NPC; in reality, it is difficult to ensure that all participants are complying with a screening programme.26 Secondly, the parameters of the model, which were sourced from

534

DP Rao, J Gu, XH Meng et al. Efficacy analysis for NPC screening


previous studies, should be updated. Although these limitations exist, the findings of the present analysis appear to have important implications for NPC screening programmes. In conclusion, the present study showed that an improved Markov model for the natural history of NPC was successfully constructed. The findings suggest that strategy B1 annual screenings for EBVseropositive subjects and biennial screenings for EBV-seronegative subjects achieved the greatest efficacy among the 16 strategies modelled. Using this strategy, the 5-year mortality rate dropped dramatically (by 42.3%) from 41.6% to 24.0% compared with no screening.

Conflicts of interest
The authors had no conflicts of interest to declare in relation to this article.

Acknowledgements
We thank Qi Hong Huang and staff at the Sihui Cancer Registry System at the Sihui Cancer Institute for their efforts in data collection. This study was supported by the 11th (2006BAI02A11; to Ming-Huang Hong) National Science and Technology Support Programme of China.

Received for publication 15 November 2011 Accepted subject to revision 20 November 2011 Revised accepted 26 January 2012 Copyright 2012 Field House Publishing LLP
References 1 Yu MC, Yuan JM: Epidemiology of nasopharyngeal carcinoma. Semin Cancer Biol 2002; 12: 421 429. 2 Xu YF, Liu WL, Dong JQ, et al: Sequencing of DC-SIGN promoter indicates an association between promoter variation and risk of nasopharyngeal carcinoma in Cantonese. BMC Med Genet 2010; 11: 161. 3 Jia WH, Pan QH, Qin HD, et al: A casecontrol and a family-based association study revealing an association between CYP2E1 polymorphisms and nasopharyngeal carcinoma risk in Cantonese. Carcinogenesis 2009; 30: 2031 2036. 4 Jia WH, Luo XY, Feng BJ, et al: Traditional Cantonese diet and nasopharyngeal carcinoma risk: a large-scale casecontrol study in Guangdong, China. BMC Cancer 2010; 10: 446. 5 Jemal A, Bray F, Center MM, et al: Global cancer statistics. CA Cancer J Clin 2011; 61: 69 90. 6 Ng WT, Yau TK, Yung RW, et al: Screening for family members of patients with nasopharyngeal carcinoma. Int J Cancer 2005; 113: 998 1001. 7 Wei WI, Sham JS: Nasopharyngeal carcinoma. Lancet 2005; 365: 2041 2054. 8 Yang XR, Diehl S, Pfeiffer R, et al: Evaluation of risk factors for nasopharyngeal carcinoma in high-risk nasopharyngeal carcinoma families in Taiwan. Cancer Epidemiol Biomarkers Prev 2005; 14: 900 905. 9 Karray H, Ayadi W, Fki L, et al: Comparison of three different serological techniques for primary diagnosis and monitoring of nasopharyngeal carcinoma in two age groups from Tunisia. J Med Virol 2005; 75: 593 602. 10 Kimura Y, Suzuki D, Tokunaga T, et al: Epidemiological analysis of nasopharyngeal carcinoma in the central region of Japan during the period from 1996 to 2005. Auris Nasus Larynx 2011; 38: 244 249. 11 National Office for Cancer Prevention and Control, National Central Cancer Registry, Disease Prevention and Control Bureau, Ministry of Health: Chinese Cancer Registry Annual Report 2009. Beijing: Military Medical Science Press, 2010; pp 284 301. 12 Mousavi SM, Sundquist J, Hemminki K: Nasopharyngeal and hypopharyngeal carcinoma risk among immigrants in Sweden. Int J Cancer 2010; 127: 2888 2892. 13 Raab-Traub N: EpsteinBarr virus in the pathogenesis of NPC. Semin Cancer Biol 2002; 12: 431 441. 14 Fachiroh J, Schouten T, Hariwiyanto B, et al: Molecular diversity of EpsteinBarr virus IgG and IgA antibody responses in nasopharyngeal carcinoma: a comparison of Indonesian, Chinese, and European subjects. J Infect Dis 2004; 190: 53 62. 15 Middeldorp JM, Brink AA, van den Brule AJ, et al: Pathogenic roles for EpsteinBarr virus (EBV) gene products in EBV-associated proliferative disorders. Crit Rev Oncol Hematol 2003; 45: 1 36. 16 Tang JW, Rohwder E, Chu IM, et al: Evaluation of EpsteinBarr virus antigen-based

535

DP Rao, J Gu, XH Meng et al. Efficacy analysis for NPC screening

immunoassays for serological diagnosis of nasopharyngeal carcinoma. J Clin Virol 2007; 40: 284 288. 17 Henle G, Henle W: EpsteinBarr virus-specific IgA serum antibodies as an outstanding feature of nasopharyngeal carcinoma. Int J Cancer 1976; 17: 1 7. 18 Lo KW, To KF, Huang DP: Focus on nasopharyngeal carcinoma. Cancer Cell 2004; 5: 423 428. 19 Hsu WL, Yu KJ, Chien YC, et al: Familial tendency and risk of nasopharyngeal carcinoma in Taiwan: effects of covariates on risk. Am J Epidemiol 2011; 173: 292 299. 20 Melbye M, Ebbesen P, Levine PH: Early primary infection and high EpsteinBarr virus antibody titers in Greenland Eskimos at high risk for nasopharyngeal carcinoma. Int J Cancer 1984; 34: 619 623. 21 Chien YC, Chen JY, Liu MY, et al: Serologic markers of EpsteinBarr virus infection and nasopharyngeal carcinoma in Taiwanese men. N Engl J Med 2001; 345: 1877 1882. 22 Zeng Y, Zhang LG, Li HY, et al: Serological mass survey for early detection of nasopharyngeal carcinoma in Wuzhou City, China. Int J Cancer 1982; 29: 139 141. 23 Zeng Y, Zhong JM, Li LY, et al: Follow-up studies on EpsteinBarr virus IgA/VCA antibodypositive persons in Zangwu County, China. Intervirology 1983; 20: 190 194. 24 Cao SM, Liu Z, Jia WH, et al: Fluctuations of EpsteinBarr virus serological antibodies and risk for nasopharyngeal carcinoma: a prospective screening study with a 20-year follow-up. PLoS One 2011; 6: e19100. 25 Ji MF, Wang DK, Yu YL, et al: Sustained elevation of EpsteinBarr virus antibody levels preceding clinical onset of nasopharyngeal carcinoma. Br J Cancer 2007; 96: 623 630. 26 Ng WT, Choi CW, Lee MC, et al: Outcomes of nasopharyngeal carcinoma screening for high risk family members in Hong Kong. Fam Cancer 2010; 9: 221 228. 27 Chen HH, Prevost TC, Duffy SW: Evaluation of screening for nasopharyngeal carcinoma: trial design using Markov chain models. Br J Cancer 1999; 79: 1894 1900. 28 Choi CW, Lee MC, Ng WT, et al: An analysis of the efficacy of serial screening for familial nasopharyngeal carcinoma based on Markov chain models. Fam Cancer 2011; 10: 133 139. 29 Fleming ID, Cooper JS, Henson DE, et al (eds): AJCC Cancer Staging Manual, 5th edn.

Philadelphia: LippincottRaven, 1997; pp 33 35. 30 Sobin LH, Wittekind (eds): TNM Classification of Malignant Tumours, 5th edn. New York: Wiley Liss, 1997; pp 25 30. 31 Kulasingam SL, Myers ER: Potential health and economic impact of adding a human papillomavirus vaccine to screening programs. JAMA 2003; 290: 781 789. 32 Newall AT, Beutels P, Wood JG, et al: Costeffectiveness analyses of human papillomavirus vaccination. Lancet Infect Dis 2007; 7: 289 296. 33 Goldie SJ, Kohli M, Grima D, et al: Projected clinical benefits and cost-effectiveness of a human papillomavirus 16/18 vaccine. J Natl Cancer Inst 2004; 96: 604 615. 34 Zeng Y, Pi GH, Deng H, et al: EpsteinBarr virus seroepidemiology in China. AIDS Res 1986; 2(suppl 1): S7 S15. 35 Jia WH, Huang QH, Liao J, et al: Trends in incidence and mortality of nasopharyngeal carcinoma over a 20 25 year period (1978/1983 2002) in Sihui and Cangwu counties in southern China. BMC Cancer 2006; 6: 178. 36 Miller DK, Homan SM: Determining transition probabilities: confusion and suggestions. Med Decis Making 1994; 14: 52 58. 37 Jiang SQ, Liu Q: Application of logistic regression in combination with multiple diagnostic tests for auxiliary diagnosis of nasopharyngeal carcinoma. Ai Zheng 2009; 28: 177 180. 38 National Office for Cancer Prevention and Control, National Central Cancer Registry, Disease Prevention and Control Bureau, Ministry of Health: Chinese Cancer Registry Annual Report 2008. Beijing: Military Medical Science Press, 2009; pp 46 49. 39 Loh KS, Goh BC, Lu J, et al: Familial nasopharyngeal carcinoma in a cohort of 200 patients. Arch Otolaryngol Head Neck Surg 2006; 132: 82 85. 40 Lee AW, Poon YF, Foo W, et al: Retrospective analysis of 5037 patients with nasopharyngeal carcinoma treated during 1976 1985: overall survival and patterns of failure. Int J Radiat Oncol Biol Phys 1992; 23: 261 270. 41 Khanfir A, Frikha M, Ghorbel A, et al: Prognostic factors in metastatic nasopharyngeal carcinoma. Cancer Radiother 2007; 11: 461 464.

Authors address for correspondence Professor Qing Liu Department of Epidemiology, Cancer Centre, Sun Yat-sen University, 651 Dongfeng Road East, Guangzhou, Guangdong 510060, China. E-mail: liuqing5@hotmail.com

536

You might also like