  |
|
|
||||
|
||||||
| Journal of the New Zealand Medical Association,
11-February-2005, Vol 118 No 1209
|
||||||
|
Elevated serum prostate-specific antigen levels and public
health implications in three New Zealand ethnic groups: European, Maori, and
Pacific Island men
Marion Gray, Barry Borman, Peter Crampton, Philip Weinstein,
Craig Wright, John Nacey
There are notable ethnic and geographic variations in the
incidence of prostate cancer.1 The highest
reported incidence is from Scandinavian countries, while there is an
intermediate incidence in America and the United Kingdom, and the lowest
incidence occurs in the Far East, especially Mainland China and
Japan.2,3
New Zealand’s population consists of three major
ethnic groups, New Zealand European, Maori, and Pacific Islands people and, for
several years, New Zealand has ranked among the six highest incidence countries
in the World.4. World Health Organization (WHO)
age-standardised prostate cancer rates in 1998–1999, showed that incidence
at 86.1 per 100,000, Maori males had the lowest incidence, followed by Pacific
Islands males at 115.2 per 100,000, and ‘other males’ (chiefly New
Zealand European) had the highest incidence (118.9 per
100,000).5 Conversely, the prostate cancer WHO
age-standardised mortality rates for Pacific Islands and Maori males in
1998–1999 (52.3 and 39.3 per 100,000 respectively) were higher than rates
for ‘other males’ (22.8 per
100,000).5
However, there are reported inaccuracies in ethnic health data collection
in New
Zealand.6
In the United States, there is a 30% higher incidence and
120% higher prostate cancer mortality rate for African-Americans, compared to
European
Americans.7
Indeed, most clinical studies indicate that African-American men have a two-fold
higher prevalence of metastatic disease at diagnosis. Differences in tumour
stage have been attributed to differences in access to and utilisation of
healthcare in African-American
men.8
Similarly, higher prostate cancer incidence among Japanese-American men
in comparison with native Japanese men is thought to be due to the more
intensive prostate cancer screening within
America.9
Healthcare access and utilisation is also an issue in New Zealand for
Maori and Pacific Islands
people10–13
and may be reflected in the difference between prostate cancer
incidence and mortality statistics shown for these groups.
When used in conjunction with a digital rectal examination,
elevated PSA levels (>4.0 ng/mL) aid in the detection of organ confined
prostate cancer.14 Specifically, it has been
found that a PSA value between 4.0–10.0 ng/mL carries a 22% probability of
prostate cancer, while a PSA value of >10.0 ng/mL increases the cancer risk
to more than 60%. PSA levels can also be raised in cases of benign prostatic
hyperplasia and prostatitis.15
Because of the lack of specificity and sensitivity, PSA
screening for asymptomatic men in New Zealand is not recommended by the National
Health Committee.16 Regardless, the risk of
prostate cancer parallels serum PSA levels and it has been suggested that
elevated PSA could be used in the identification of high prostate cancer risk
cohorts.17
A Japanese study (Shibata et al) compared the numbers of undetected prostate
cancers between two populations, using the prevalence of elevated
PSA.9 Shibata et al showed that even though
prostate cancer incidence is currently four-fold to six-fold higher in
Japanese-American men than native Japanese, it is likely to be lower (1.9 times)
when undetected cancers are accounted
for.9
Determining the prevalence of elevated PSA within New
Zealand's major ethnic groups may provide some indication of the extent to which
inadequate access to and under-utilisation of healthcare services influences New
Zealand prostate cancer incidence statistics.
In the course of examining PSA levels in a non-random
population with no clinical history of prostate cancer, we were able to evaluate
the cross-sectional prevalence of elevated PSA at enrolment. Our study provides
estimates of elevated PSA prevalence, prostate cancer incidence, and predicted
cancer prevalence ratios among the three major New Zealand ethnic groups.
MethodsSubjects—Wellington
Ethics Committee approval was gained. Written informed consent was obtained from
each subject. Between January 2000 and February 2002 inclusive, a total of 1617
males were recruited into the Wellington Regional Community Prostate Study
(Wellington Study), coordinated through the Wellington School of Medicine and
Health Sciences of the University of Otago. To ensure an adequate representation
of Maori and Pacific Islands men, subjects were non-randomly enrolled by two
separate means. Initially males aged between 40 and 69 years (within selected
census area units containing at least 5% Maori and 5% Pacific Islands
populations) were invited to attend a local clinic. A blood sample was taken and
a detailed questionnaire was completed (Phase 1). The total number recruited
into Phase 1 was 698.
The second mode of recruitment was through invitation
to all eligible individuals who had been screened as part of the Wellington
hepatitis and diabetes-screening programme for Maori and Pacific Islands
populations. Upon recruitment, blood samples were retrieved from the Hepatitis
Foundation of New Zealand and subjects completed the study questionnaire (Phase
2). The total number recruited into Phase 2 was 919.
Questionnaires utilised tick boxes and were
self-administered. Ethnicity was determined on a self-identification basis (as
in the 1986 New Zealand Census of Population and Dwellings ethnicity question;
NZHIS 2001). Subjects were categorised as Maori or Pacific Islands if they
indicated this affiliation as well as other ethnic affiliations. Each respondent
was asked to grade the severity of urinary symptoms (using the International
Prostate Symptom Score),18 and to declare if
they have ever had any evidence of prostate disease. Subjects with total PSA
levels of >4.0 ng/mL were referred back to their general practitioner (GP)
for further evaluation for prostatic malignancy.
To ensure subjects were affiliated to New Zealand
European or Maori or Pacific Islands groups and aged between 40-69, age and
ethnicity selection criteria were applied to the combined Phase 1 and Phase 2
group (1617). Twenty-eight subjects were excluded on the basis of indicating
Asian affiliation, and three subjects were excluded on the basis of prior
prostate cancer. There were 1425 eligible subjects—728 New Zealand
European, 353 Maori, and 344 Pacific Islands men.
A literature review undertaken during the planning of
this study demonstrated that in population-based studies across all ages
(40–90 years), 9%–13% of subjects had PSA levels elevated above 4.0
ng/mL.8,14,19,20
The power calculation determined a sample size of 343 per ethnic group, assuming
elevated PSA levels would occur in 9%–13% of subjects and allowing an 80%
chance of finding a significant difference at the p<0.05 level in a t-test
comparing geometric means between ethnic groups.
Sera from each subject were tested for total PSA and
free PSA. A 10mL tube of blood was collected, and centrifuged within 4 hours of
sampling. Serum was stored at -70°C until assayed.
Laboratory blood analysis for PSA was carried out using
an Elecsys 2010 assay (Roche Diagnostics, Mannheim, Germany). Assays were
performed according to manufacturer's specifications.
Statistical
analyses—Statistical analyses were performed with SPSS version 10.1
for the PC (SPSS Inc, Chicago, IL). Two separate sets of analyses were
developed, first to investigate the overall and ethnic-specific prevalence of
elevated PSA, as based on the currently used cut-off of over 4.0
ng/mL,21 and second to examine the use of
elevated PSA as a means of estimating potential levels of undetected cancer and
prostate cancer burden in New Zealand's ethnic groups.
Prevalence estimates for the larger Wellington target
population were calculated by standardising by population proportions using the
most recent Census (2001)
data.22
Chi-squared tests were
performed to determine whether the prevalence of elevated PSA varied
significantly by ethnicity. A logistic regression model was developed to test
whether the prevalence of elevated PSA was associated with a number of measured
demographic and clinical characteristics, including ethnicity and socioeconomic
position (using
NZDep96).23
Comparison of age structure of elevated PSA by ethnic group was made using an
interaction term (ethnicity * age) in the model. Tests of statistical
significance were two-sided.
NZDep96 provides a deprivation score for each meshblock
in New Zealand. Meshblocks are geographical units defined by Statistics New
Zealand, containing a median of 90 people. The NZDep96 scale runs from 1 to 10
so that a value of 10 indicates that the meshblock is in the most deprived 10%
of areas in New
Zealand.23
1998-1999 prostate cancer registration data from the
New Zealand Health Information Service were utilised. Comparisons of levels of
elevated PSA and prostate cancer incidence, as well as predictions of prostate
cancer detection rates, were undertaken by graphically modelling crude
age-specific prostate cancer incidence rates and prevalences of elevated PSA by
ethnic group in New Zealand and other selected countries. The proportion of
prostate cancers in relation to the prevalence of elevated PSA was calculated.
Shibata et al’s methodology provided an estimate
of cancer burden based on elevated PSA and was used as a guide for the analysis
of cancer prevalence ratios between New Zealand's three main ethnic groups. We
initially followed Shibata et al’s method, which used age-specific
reference ranges, based on the upper PSA cut-off level for each age 10-year
group in their study population, to determine the age-specific prevalence of
elevated PSA (P) in two populations.9 After
this point, we modified Shibata et al’s original methodology (see Appendix
1).
The utilisation of total PSA as a predictor of
malignancy may be enhanced by the use of age-specific reference ranges, which
increase the sensitivity of PSA screening in younger males and the specificity
in older males.24 The age-specific reference
range used to determine the prevalence of elevated PSA in the Wellington Study
group was calculated using a best-fit regression of the
95th percentile cut-offs, devised on the
subjects showing no evidence of clinical prostate
cancer.25
We estimated the prevalence of undetected cancer in all
three New Zealand ethnic groups with elevated PSA, by using the age-specific
positive predictive value (PPV) based on the age-specific reference range for
elevated PSA from a large population of American European
volunteers.14
Estimates of undetected cancer for each ethnic and
10-year age group were calculated by the following formula:
The
estimated proportion of undetected cancers = the expected PPV x (number with
elevated PSA/total number of subjects) x 100.
To determine the crude prostate cancer incidence rates
that coincided most closely with the Wellington Study’s timeframe, ethnic
and age-specific incidence rates were calculated for men aged between
40–69 years, for the years 1998–1999. To avoid the possible
inclusion of study participants who were found to have prostate cancer, prostate
cancer registration data after this date were not included.
Estimates of the prevalence of undetected cancer were
combined with the incidence of detected cancer to find the cumulative risk of
having an undetected cancer. These risk estimates provided estimates of the
total cancer
burden.9
The fraction of undiagnosed cancers was estimated by
the formula:9
%
undiagnosed cancer = 100 - (%detected cancers / %cumulative risk of undetected
cancer) x 100.
Predicted cancer prevalence ratios were calculated
using the % cumulative risks.9 For example;
Predicted
cancer prevalence ratio for New Zealand European compared with Maori =
%cumulative risk New Zealand European / %cumulative risk Maori.
Cancer incidence ratios were calculated on 1998-99
prostate cancer registration statistics. For
example;
Cancer
incidence ratio for New Zealand European compared with Maori = New Zealand
European age-specific cancer incidence rate/Maori age-specific cancer incidence
rate.
The predicted cancer prevalence ratios were compared
with cancer incidence
ratios.9
The PPV is likely to differ by ethnicity, as the risk
of getting cancer at elevated PSA levels is different between ethnic
groups.8,21
We explored the effect of an ethnic difference in PPV on results by
recalculating predicted cancer prevalence ratios. For example, we applied 18.2%
PPV for Maori men (as for Japanese men; to illustrate an extreme low
PPV)26 and had New Zealand European's PPV
remain at 31.8%.
ResultsTable 1 gives the New Zealand
prevalence of elevated PSA compared to other ethnic groups. It shows that the
age-standardised elevated PSA prevalences for New Zealand European, Maori and
Pacific Islands men (3.7, 3.3, and 4.2% respectively) are lower than all other
community-based studies
show.8,14,26–29
The chi-squared tests confirmed that the
difference in the prevalence of elevated PSA between the New Zealand ethnic
groups and the other groups was statistically significant (p<0.05).
Figure 1 is a comparison of crude age-specific prostate
cancer incidence rates and the prevalence of raised PSA by ethnic group in New
Zealand and other selected countries. Although the prevalence of elevated PSA is
substantially higher than the prostate cancer incidence rates, ethnic
differences in levels of elevated PSA generally reflect the trends in difference
in incidence between ethnic groups.
New Zealand European men (at 5.2%) have a higher proportion
of prostate cancers in relation to elevated PSA levels than all other groups,
followed by African-Americans (2.3%), Pacific Islands (2.2%), Maori (2.1%), and
American Europeans (1.9%). Japanese men appear to have lower incidence rates in
comparison to elevated PSA levels, than shown for all other ethnic groups
(0.2%).
Table 2 gives the percentage of men in the study population
with a PSA value above the current reference range (0.0–4.0 ng/mL),
compared to the age-specific reference ranges (as used for calculating the
prevalence of undetected cancer), by age group.
The prevalence of elevated PSA in each age and ethnic group
increased using age-specific ranges, with the exception of Maori men aged
60–69 years, for whom the prevalence decreased (13.2% using 0.0–4.0
ng/mL; compared to 7.5% using age-specific reference ranges).
Only age was significantly related to the chances of having
an elevated PSA level as defined by either the current reference range
(0.0–4.0 ng/mL) or the age-specific range (p<0.001). The chi-squared
test and logistic regression model demonstrated no significant difference
between the New Zealand ethnic groups in the prevalence of elevated PSA and no
differences in the chances of getting a change in elevated PSA age pattern
between the ethnic groups. NZDep96 was not associated with the prevalence of
elevated PSA. Potential confounding factors were adjusted for within the
multivariate models.
Table 3 gives the estimated prevalence of undetected
prostate cancer by age and ethnic group. Data show that 2.2% of Pacific Islands
and New Zealand European men aged 40–69 have undetected prostate cancer
and 2.1% of Maori men have undetected prostate cancer. While, overall, New
Zealand European men have the highest cumulative risk of prostate cancer (2.4%),
Maori men have the highest cumulative risk in the 40-49 year age group (1.4%)
and Pacific Islands men have the highest cumulative risk in the 50-59 and 60-69
year age groups (2.1 and 4.4% respectively).
Calculations suggest that Pacific Islands men have the
highest fraction of prostate cancer (95.6%) remaining undiagnosed of all ethnic
groups. With Maori intermediate at 95.5% undiagnosed and New Zealand Europeans
the lowest fraction undiagnosed (91.7%).
Table 4 gives the cancer incidence ratios and predicted
cancer prevalence ratios between New Zealand European, Maori and Pacific Islands
men. Prostate cancer registration data from 1998 to 1999 shows the cancer
incidence ratios between New Zealand Europeans compared with Maori is 2.7, New
Zealand European compared with Pacific Islands men is 2.0, and Pacific Islands
compared with Maori men is 1.3. However, the predicted cancer prevalence ratios
are 1.1 across all ethnic comparisons.
ConclusionsWe compared ethnic groups in
Wellington (New Zealand) and Japan, Austria, United States, Singapore, and the
Caribbean region using >4.0 ng/mL as a cut-off for determining levels of
elevated PSA. Although there was no significant difference in the prevalence of
elevated PSA between the New Zealand ethnic groups, there was a difference
between the New Zealand groups and the prevalences shown for ethnic groups in
other community-based studies.
The overall New Zealand age-standardised elevated PSA
prevalence (3.9%) was closest to the prevalence found for Japanese men
(7.8%)26 and lower than the prevalence of any
other ethnic group shown (Table 1). For example, Austrian European
(8%),29 American European
(9.7%),14 African-American
(13%),8 Singaporean Asian
(13.1%),28 and Afro-Caribbean
(31%).
27
Available New Zealand prostate cancer crude incidence rates
showed that, among different ethnic groups, New Zealand Europeans had the
highest rate, followed by Pacific Islands men and lowest were Maori men (191,
94.2, and 70.4 per 100,000 respectively). In general, New Zealand men were most
similar to American European men in age-specific prostate cancer crude incidence
rates.
The prevalence of elevated PSA reflects ethnic differences
in prostate cancer incidence (Figure 1). Results highlight that levels of
elevated PSA can parallel prostate cancer
risk17—and thus may be useful in
estimating the incidence of undetected cancers in a population. For example,
data suggests that African-Americans have both the highest prostate cancer
incidence and prevalence of elevated PSA; equally, Japanese men have both the
lowest prostate cancer incidence and prevalence of elevated PSA.
Determining elevated PSA using age-specific reference ranges
generally increased the elevated PSA prevalence estimates, indicating a possible
degree of under-diagnosis in all New Zealand groups through PSA based screening
using the standard cut-off (>4.0 ng/mL) (Table 2). The use of an
age-specific PSA reference range has been found to produce more accurate
results.19
Predicted cancer risks for New Zealand ethnic groups were
age-dependant. As younger men (under 50 years old) who develop this disease are
less likely to survive it,2 age-specific
prostate cancer risk could be important. Maori men were shown to have the
highest cumulative risk when younger (40–49 years), whereas Pacific
Islands men had higher risks when aged between 50–69 (Table 3).
Results suggest that New Zealand Europeans had the highest
rate of prostate cancer detection of any other ethnic group compared and that
estimates of undiagnosed cancer were highest for Pacific Islands men. Data also
show that at over two-fold the incidence, prostate cancer incidence ratios based
on available New Zealand prostate cancer registration data likely overestimate
differences in incidence rates between New Zealand European, and Maori and
Pacific Islands men. The incidence ratios are more likely to be closer to 1.
Under-utilisation of health services by Maori and Pacific Islands men and
under-reporting of ethnicity in health data, would lead to fewer prostate cancer
diagnoses, and thus, lower incidence statistics shown for these groups.
Statistics show that although more New Zealand European men
are diagnosed with prostate cancer, more Pacific Islands and Maori men die from
this disease. This supports the possibility that Pacific Islands and Maori men
receive prostate health care later than other ethnic groups, as has been found
for African
Americans.
8 Cultural barriers to healthcare access are likely to be an important
factor determining prostate healthcare utilisation rates amongst Maori and
Pacific Islands
people.10–13,30
There are several caveats to consider with our study.
Elevations of serum PSA levels beyond that accounted for by prostatic size
alone, may reflect levels of non-detectable prostate
cancer.21
As there was no difference in the prevalence of significant lower urinary
tract symptoms between the three Wellington Study’s ethnic groups, we
believe that levels of elevated PSA in subjects may reflect ethnic-specific
levels of non-detectable prostate cancer.
Estimates of undetected cancer prevalence only include
cancers that cause an elevated PSA level, whereas the detected cancer prevalence
rates include cancers that are still too small to do so. Therefore, there may be
some underestimation in undetected prostate cancer incidence using elevated PSA
alone.9
The potential for self-selection bias in our non-random
sample was quantitatively evaluated using a variety of methods, including: the
exclusion of groups with possible biases such as urinary symptoms; comparison of
urinary symptoms with a randomised New Zealand
study;30 standardisation for age, smoking and
NZDep96 and sensitivity analyses. All investigations indicated that
self-selection bias was unlikely to explain levels of elevated PSA in our
Wellington Study.
In conclusion, the prevalence of elevated PSA in New Zealand
men was lower than found in other community-based studies and not significantly
different between the three New Zealand ethnic groups. However, levels of
elevated PSA may be useful for predicting prostate cancer incidence rates in
ethnic groups.
Available incidence data show New Zealand European men to
have a higher prostate cancer incidence rate than both Maori and Pacific Islands
men; however, this study found that prostate cancer incidence ratios between
these groups are more likely to be closer to 1. Therefore, prostate cancer
incidence appears to be at least as high in Maori and Pacific Islands men as New
Zealand European men, and prostate cancer related mortality greater.
It is likely that under-utilisation of health services by
Maori and Pacific Islands men is reflected in the lower prostate cancer
incidence shown for these groups. Findings may indicate cultural barriers in the
health system for Maori and Pacific Islands men; highlighting the need for
clinicians to further consider cultural appropriateness in practice and target
prostate health promotion for these groups.
Appendix
1: Discussion of methodology—Shibata
et al estimated the prevalence of undetected cancers in the participants with
elevated PSA, using a formula that determines that the excess prevalence minus
the false positive rate (P-0.05) is approximately the proportion of men with
undetected cancer. However, this method assumes that the levels of false
positives will be consistent across the age
groups.9
Such consistency across age groups has been shown to be
untrue.14
As a significant parameter in
determining the value of cancer detection tests is the
PPV,14 we used the PPV for the purposes of our
research.
The PPV is the fraction of
patients who have a cancer when a method of detection, such as PSA screening,
shows a positive
result.14
Because not all study subjects were referred for further prostate cancer
diagnoses and data were not available for all those who attended their GP, we
could not calculate PPVs for our Wellington Study Group.
Therefore, to estimate the
numbers of undetected cancers in our Wellington Study, New Zealand incidence
data were compared with overseas incidence data to establish the most
appropriate PPV. We used a PPV based on the age-specific reference range for
elevated PSA from a large population of American European
volunteers, because their prostate cancer
incidence rates were the closest to all three ethnic groups in our
study.14 Further research into actual PPVs for
New Zealand populations could make such a method more useful.
Because of the need to assume
a PPV for ethnic groups within the Wellington Study there are important
limitations to this method of estimating cancer incidence. If Maori men's PPV
were 18.2% and New Zealand Europeans remained at 31.8%, the estimated number of
undetected cancers overall in Maori men would decrease (from 2.1% to 1.2%), as
would the cumulative risk of prostate cancer (from 2.2% to 1.3%). This, in turn
would cause an increase in estimated cancer prevalence ratio of New Zealand
European men compared to Maori men (from 1.1 to 1.9). As the resulting estimated
cancer prevalence (1.9) is still lower than that shown by available incidence
statistics (2.7), we feel that the potential for ethnic differences in PPV do
not fully explain our results.
Author information:
Marion A Gray, Phd Candidate, Department of
Public Health, Wellington School of Medicine and Health Sciences, University of
Otago, Wellington; Barry Borman, Manager, Public Health Intelligence, Ministry
of Health, Wellington; Peter Crampton, Head of Department, Department of
Public Health, Wellington School of Medicine and Health Sciences, University of
Otago, Wellington; Philip Weinstein, Professor,
School of Population Health, The University of Western Australia, Perth,
Australia; Craig S Wright, Advisor (statistics), Public Health Intelligence,
Ministry of Health, Wellington; John N Nacey, Dean, Wellington School of
Medicine and Health Sciences, University of Otago, Wellington
Acknowledgements
This project was supported by funds from the Wellington School of Medicine
Surgical Research Trust, Community Trust of Wellington, and University of Otago.
We would also like to acknowledge the support of The Hepatitis Foundation of New
Zealand and statistical input from Mr Gordon Purdie, Statistician, Wellington
School of Medicine and Health Sciences.
Correspondence Dr
Marion A Gray, Armed Forces Institute of Pathology, Dept of Environmental and
Infectious Disease Sciences, 6825 16th St N.W., Bldg 54, Room M098, Washington,
DC 20306-6000, United States. Fax: (202) 782 9215, email: marion.gray@afip.osd.mil
References:
|
|
||||
| | Current
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads) Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals |
|