Draft Genetic Test Review

---

Cystic Fibrosis
Clinical Validity

(436KB)

---

CLINICAL VALIDITY

Question 18: How often is the test positive when the disorder is present?
Question 19: How often is the test negative when the disorder is not present?
Question 20: Are there methods to resolve clinical false positive results in a timely manner?
Question 21: What is the prevalence of the disorder in this setting?
Question 22: Has the test been adequately validated on all populations to which it may be offered?
Question 23: What are the positive and negative predictive values?
Question 24: What are the genotype/phenotype relationships?
Question 25: What are the genetic, environmental or other modifiers?

---

CLINICAL VALIDITY

Question 21: What is the birth prevalence of cystic fibrosis in the prenatal setting?

page 1 | page 2 | page 3

Introduction
In the prenatal screening setting, the birth prevalence of a given disorder is defined as the number of cases that would be present at birth in the absence of prenatal diagnosis and selective termination. The birth prevalence of cystic fibrosis can be estimated from three separate data sources:

• prenatal screening trials,
• newborn screening studies,
• population-based studies and registries.

When analyzing prevalence, it is important to consider both geographical and racial/ethnic variability. The first analysis presented is restricted to non-Hispanic Caucasians of Northern or Southern European heritage, since the majority of published data deals with this group. Later, similar analyses will be provided for other groups that include:

• Ashkenazi Jewish Caucasians
• Hispanic Caucasians
• African Americans/Afro-Caribbean Blacks
• Asian Americans

Cystic fibrosis prevalence: Prenatal screening trials
Thirteen population-based prenatal screening trials have been published that can be used to estimate the prevalence of cystic fibrosis among non-Hispanic Caucasians of Northern European heritage. The estimate relies on the observed carrier rate among pregnant women being tested for a cystic fibrosis mutation (without a family history of cystic fibrosis), in combination with each individual trial's mutation detection rate. For example, a trial might report 33 carrier women identified among 1000 women tested (observed carrier rate of 0.033). That trial also reports that its laboratory uses a mutation panel that identifies 80 percent of all cystic fibrosis mutations. The carrier rate can then be corrected to 0.04125 (0.033/0.8), to take this into account. This, in turn, allows the prevalence of cystic fibrosis to be estimated at 0.0004254 (1/4 * 0.04125 * 0.04125) or about 1:2350.

Table 3-19 shows, for each of the 13 studies, the number of individuals (women and their partners) tested, the number of carriers detected, the observed carrier rate, the mutation detection rate, and the corrected carrier rate. The computed prevalence and 95% confidence intervals are listed in the last two columns. Overall, a cystic fibrosis mutation is detected in 1,233 of the 39,284 individuals tested.

Table 3-19. Estimated Cystic Fibrosis Prevalence Derived from Prenatal Screening Trials

Study Number	Persons Tested	Carriers Detected	Observed Rate	Mutations Detected (%)	Corrected Rate	CF Prevalence (1 in n)	95% CI
1	3,275	115	0.0351	85.0	0.0413	2344	1694-3322
2	6,761	175	0.0259	87.7	0.0295	4592	3522-6151
3	1,167	39	0.0303	84.5	0.0359	3110	1780-5783
4	562	19	0.0338	80.0	0.0423	2240	1071-5739
5	658	18	0.0274	73.2	0.0374	2864	1308-6719
6	4,413	160	0.0363	83.0	0.0437	2090	1606-2829
7	1,867	62	0.0332	91.0	0.0365	3004	1899-4959
8	4,210	108	0.0257	75.1	0.0342	3428	2488-4789
9A	1,091	32	0.0293	75.0	0.0391	2615	1451-4925
9B	2,633	97	0.0368	80.0	0.0461	1886	1337-2727
10	3,948	135	0.0342	80.0	0.0427	2189	1613-2970
11	3,286	74	0.0225	65.0	0.0346	3332	2323-4856
12	1,621	47	0.0290	75.0	0.0387	2676	1632-4444
13	3,792	152	0.0401	96.0	0.0418	2294	1700-3174
Total	39,284	1233

The following is a brief discussion of how the numbers were obtained from each study listed in the table. For additional information about the outcomes of these screening trials, see Pilot Trials (Clinical Utility, Question 33).

1. Edinburgh, Scotland – Only the first report (Mennie et al., 1992) is included in the table. In later reports, it is not possible to separate women with no family history from the relatively large number of women with a known family history. The population is assumed to be entirely Scottish; several potential participants were not screened, due to low prevalence in their racial/ethnic group. The mutation detection rate of 85 percent was derived from a local population (Shrimpton et al., 1991). Overall, 3,165 women (and 110 partners) were tested with 111 carrier women (and four carrier partners) detected.
2. Copenhagen, Denmark – All data from this report (Schwartz et al., 1993) are included in the table. The Scandinavian countries are known to have a lower prevalence of cystic fibrosis, and this is borne out in the estimated prevalence of 1:4529, by far the lowest of the 13 studies. The mutation detection rate was derived from a local population and is contained in the original reference. None of the women tested had a family history of cystic fibrosis. Overall 6,599 women (and 162 partners) were tested, with 172 carrier women and three carrier partners identified.
3. Manchester, England – The data were derived from two reports of this screening program (Harris et al., 1993; Hartley et al., 1997). Couples with a family history were allowed into the study but were not actively recruited. The mutation detection rate was derived from an unpublished study in a local population. In the first report, 127 women (and five partners) were tested, with five carrier women detected. In addition, 117 women (and all 117 partners) were tested, with a total of eight carrier individuals detected. In the second report, 267 women (and 10 partners) were tested, with 10 carrier women detected, along with 262 women (and all 262 partners), with a total of 16 carrier individuals detected.
4. Oxford, England – The data were derived from two reports of this screening program (Wald et al., 1993; Wald et al., 1995). It is assumed that couples with a family history were allowed into the study but were not actively recruited. No reference was quoted for the mutation detection rate of 80 percent, but this estimate is similar to the 80.1 percent summary estimate published for the United Kingdom (Dequeker et al., 2000). The numbers include samples tested from the male partner after a mutation was initially identified in the woman. Using the unduplicated numbers from the second report, 543 women (and 19 partners) were tested, with 19 carrier women detected.
5. East Berlin, Germany – The data were derived from a single report (Jung et al., 1993). It is assumed that couples with a family history were allowed into the study but were not actively recruited. No reference was quoted for the mutation detection rate of 61.2 percent, which is much lower than the 73.2 percent published summary estimate for Germany (Dequeker et al., 2000). The present analysis uses the 73.2 percent estimate. Overall, 637 women and 3 men were initially tested (and 18 partners), with a total of 18 women carriers and one partner carrier identified.
6. Maine, USA – The data were derived from two published studies (Doherty et al., 1996; Bradley et al., 1998) and a personal communication (Bradley, 2000). Couples with a known family history were not included. The population was 99 percent Caucasian and mainly of Northern European heritage. The mutation detection rate of 80 percent was based on data from the Cystic Fibrosis Genetic Analysis Consortium (1994). Overall, 4,260 women were tested (along with 153 partners), with a total of 153 women carriers and 7 partner carriers detected.
7. Aberdeen, Scotland – The data were derived from one published study (Miedzybrodzka et al., 1995). The study included 12 couples with a known family history of cystic fibrosis; no adjustment is made for this high risk group. The mutation detection rate of 92 percent was derived from a local population (Miedzybrodzka et al, 1993). In the two-step arm of the study, 1,487 women were tested (along with 47 partners), with a total of 48 women carriers and one partner carrier detected. In the one-step arm of the study, 321 women were tested (along with 12 partners), with a total of 12 women carriers and one partner carrier detected.
8. Rochester, New York – The data were derived from one published study (Loader et al, 1996) and a personal communication (Rowley, 1998). Of the 4,879 couples successfully tested, only 4,391 were pregnant, and 109 carrier women were detected. A family history was identified in 27 participants, 4 of whom were also carriers. All 27 were removed from this analysis. A total of 96 partners were tested, and 5 carriers were detected. The population was 94.4 percent non-Hispanic Caucasian. The estimated 250 participants remaining were also removed from the analysis. It was estimated that two mutations occurred in this group. Thus, an estimated 4,210 non-Hispanic Caucasian couples without a known family history were tested, with 108 carrier individuals identified. The reported mutation detection rate of 75.1 percent was based on the Cystic Fibrosis Genetic Analysis Consortium (1994).
9. Northern California, USA – The data were derived from a single publication (Witt et al., 1996) and a personal communication (Witt, 1998). Laboratory analysis was performed in two laboratories (A and B) with different mutation detection rates. It is for this reason that two separate results are computed for this trial. In Laboratory A, 1,091 non-Hispanic Caucasian women were tested, and 32 carrier women were identified. In Laboratory B, 2,633 non-Hispanic Caucasian women were tested, and 95 carrier women and one compound heterozygote were identified. The test results in the partners were not stratified by race and were, therefore, not included. The authors estimated mutation detection rates of 80 and 85 percent for Laboratories A and B. However, these estimates have been adjusted to 75 and 80 percent, respectively, to makes these rates consistent with those used in other pilot studies and to reflect the large proportion of non-Hispanic Caucasians in California (e.g., Italians, Greeks) who are not of northern European heritage compared to other less diverse populations (e.g., Maine).
10. Leeds, England –The data were derived from a single report (Cuckle et al., 1996). It is assumed that couples with a family history were allowed into the study but were not actively recruited. The mutation detection rate was reported to be between 80 and 90 percent in Yorkshire, but no reference as provided. The present analysis uses 85 percent. Overall, 3,773 women (and 127 partners) were tested, and 130 carrier women and 3 carrier partners were identified. In addition, 48 men attending with their partner requested immediate testing, and two carriers were identified.
11. Milan, Italy – The data were derived from a single report (Brambati et al., 1996). Among 2,231 parents without a family history of cystic fibrosis, 46 carrier individuals were identified. In addition, 1,055 fetuses were tested, and 26 carriers and one compound heterozygote were identified. The mutation detection rate was reported to be 65 percent, based on an unpublished report. This is consistent with the 65.2 percent published summary estimate for Italy (Dequeker et al., 2000).
12. Los Angeles, California– The data were derived from a single report (Grody et al., 1997) and a personal communication (Grody, 1998). Couples with a family history were excluded. The reported mutation detection rate is 75 percent. In this study, 47 carrier women were identified among 1,851 non-Hispanic Caucasian women tested. However, interpretation of these results is complicated by the fact that some ethnic groups were counted twice (i.e., the sum of the reported ethnic-specific prevalences reported is 270 observations higher than the total number of tests performed). In order to provide a reasonably reliable estimate, we have assumed that the Hispanic and non-Hispanic Caucasian groups accounted for 90 percent of the double-counting (Table 4 in Grody et al., 1997). The double-counts are then divided between the two groups in the ratio of Hispanic to non-Hispanic study subjects. Thus, 160 ([1596 / (1596+921)) * 270 * .9] observations are subtracted from the denominator of 1,851, leaving 1,621 women tested. Among this group of non-Hispanic Caucasians, 365 (23 percent) are Ashkenazi Jewish.
13. New York City, USA – The data were derived from a single report (Eng et al., 1997). The population was Ashkenazi Jewish Caucasians without a family history of cystic fibrosis. The mutation detection rate of 96 percent was from a published source specific to that population group. Overall, 3,792 individuals were tested, with 152 carrier individuals identified.

Except for the study from Denmark, the remaining 12 studies (13 observations) are reasonably consistent in their estimates of cystic fibrosis prevalence, even . Figure 3-7 shows the results of a formal meta-analysis (Berlin et al., 1989). The X-axis is labeled with the study number, in increasing order from lowest to highest estimate of prevalence. The Y-axis shows the prevalence estimates (circles), along with the 95% confidence interval (vertical lines). The horizontal solid line shows the consensus estimate, with the corresponding 95% confidence interval shown by broken horizontal lines. Overall, the prevalence is estimated to be 1:2488 (95 percent CI 1:2224 to 1:2782). A formal test for heterogeneity is statistically significant (χ² = 12, p < 0.001), indicating that the between-study differences in prevalence estimates are unlikely to have occurred by chance. When stratified by whether or not the studies are from the United States (5 studies, 6 observations) or from Europe (7 studies), the two estimates are still similar: 1:2403, (95 percent CI 1:1995 to 1:2894) and 1:2577 (95 percent CI 1:2192 to 1:3029), respectively. Heterogeneity is reduced, but still significant (χ² = 8, p = 0.005 and χ² = 4, p =0.04).

Figure 3-7 – Summary of Cystic Fibrosis Prevalence Estimates Derived From Prenatal Screening Trials

Strengths and weaknesses of these estimates

Family history of cystic fibrosis Most studies included only couples without a family history. These studies will tend to underestimate the prevalence of cystic fibrosis, but only by a small amount (about 2 percent). A few additional studies were corrected for the inclusion of couples with a family history, and the remainder of the studies allowed those couples to be screened. Because of this mix, the estimates provided are not likely to be significantly biased with respect to the inclusion, or exclusion, of couples with a family history of cystic fibrosis.

Using carrier rates to estimate prevalence The use of carrier rates to estimate prevalence of cystic fibrosis is reliable, if the mutations detected have high penetrance (i.e., individuals with two of the detectable mutations will have the cystic fibrosis phenotype). The majority of studies relied on relatively small mutational panels consisting of well-described mutations that are known to cause cystic fibrosis. It is important to note, however, that one study routinely tested for a less common mutation (R117H) that is known to be associated with non-cystic fibrosis phenotypes. In that study, the R117H mutation accounted for 16 percent of the carriers. In a population of individuals with cystic fibrosis, this mutation accounts for less than 1 percent of the mutations. Thus, among laboratories testing for the less common and less well described mutations, an artificially high prevalence might be anticipated. One advantage of using this method to indirectly estimate the prevalence is that many fewer individuals need to be tested. The largest study included less than 7,000 couples. In a group that size, only two or three affected fetuses would be expected. However, when carrier rates are used to estimate prevalence, even the studies of 2,000 couples yielded individually reliable estimates.

A note on the demographic identifiers used throughout this document We realize that demographic identifiers of race and ethnicity are potentially controversial. The biological validity of race has been challenged by biologists, and the original definition of ethnicity is a category that would describe non-biological, but socially meaningful, groups. In this document, we follow the usage of the investigators whose studies we discuss. In all studies from the United States, those categories follow current Census Bureau usage. The major, relatively new element of the Census Bureau demographic categories is the addition of Hispanic as a separate category and the separation of Hispanic into "white" and "non-white" categories. While we recognize potential errors with the use of these categories, only one presents hazards for the topic at hand. That is the extrapolation of rates of cystic fibrosis disease and gene prevalence to populations which are biologically dissimilar from those in a cited study. For example, Hispanic whites in Florida could have different rates and types of mutations for cystic fibrosis from the indigenous groups in Mexico and Central America. The only remedy for this problem is to keep a high level of awareness of the potential for genetic differences between groups that would currently fall under the same rubric of demographic identification. Data on frequency and type of mutations, if collected and analyzed as part of prenatal screening for cystic fibrosis, will allow more refined estimates to be made in the future.

References

Berlin JA, Laird NM, Sacks HS, Chalmers T. 1989. A comparison of statistical methods for combining event rates from clinical trials. Stat Med 8:141-151.

Bradley LA; Johnson DD; Doherty RA; Palomaki GE; Haddow JE. 1998. Routine prenatal cystic fibrosis screening in primary care offices. Am J Hum Genet 63:A13.

Brambati B, Anelli MC, Tului L. 1996. Prenatal cystic fibrosis screening in a low-risk population undergoing chorionic villus sampling for fetal karyotyping. Clin Genet 50:23-27.

Brock DJH. 1996. Prenatal screening for cystic fibrosis: 5 years' experience reviewed. Lancet 347:148-150.

Cuckle H, Quirke P, Sehmi I, Lewis F, Murray J, Cross D, Cuckle P, Ozols B. 1996. Antenatal screening for cystic fibrosis. Br J Obstet Gynaecol 103:795-799. >

Doherty RA, Palomaki GE, Kloza EM, Erickson JL, Dostal DA, Haddow JE. 1994. Prenatal screening for cystic fibrosis. Lancet 343:172.

Doherty RA, Palomaki GE, Kloza EM, Erickson JL, Haddow JE. 1996. Couple-based prenatal screening for cystic fibrosis in primary care settings. Prenat Diagn 16:397-404.

Eng CM, Schechter C, Robinowitz J, Fulop G, Burgert T, Levy B, Zinberg R, Desnick RJ. 1997. Prenatal genetic carrier testing using triple disease screening. JAMA 278:1268-1272.

Grody WW, Dunkel-Schetter C, Tatsugawa ZH, Fox MA, Fang CY, Cantor RM, et al. 1997. PCR-based screening for cystic fibrosis carrier mutations in an ethnically diverse pregnant population. Am J Hum Genet 60:935-947.

Hartley NE, Scotcher D, Harris H, Williamson P, Wallace A, Craufurd D, et al. 1997. The uptake and acceptability to patients of cystic fibrosis carrier testing offered in pregnancy by the GP. J Med Genet 34:459-464.

Holloway S, Brock DJH. 1994. Cascade testing for the identification of carriers of cystic fibrosis. J Med Screen 1:159-164.

Jung U, Urner U, Grade K, Coutelle C. 1994. Acceptability of carrier screening for cystic fibrosis during pregnancy in a German population. Hum Genet 94:19-24.

Livingstone J, Axton RA, Gilfillan A, Mennie M, Compton M, Liston WA, et al. 1994. Antenatal screening for cystic fibrosis: a trial of the couple model. BMJ 308:1459-1462.

Loader S, Caldwell P, Kozyra A, Levenkron JC, Boehm CD, Kazazian HH, et al. 1996. Cystic fibrosis carrier population screening in the primary care setting. Am J Hum Genet 59:234-247.

Mennie ME, Gilfillan A, Compton M, Curtis L, Liston WA, Pullen I, et al. 1992. Prenatal screening for cystic fibrosis. Lancet 340:214-216.

Miedzybrodzka ZH, Dan JCS, Russel G, Friend JAR, Kelly KF, Haites NB. 1993. Prevalence of cystic fibrosis mutations in the Grampain region of Scotland. J Med Genet 30:316-317.

Miedzybrodzka ZH, Hall MH, Mollison J, Templeton A, Russell IT, Dean JCS, et al. 1995. Antenatal screening for carriers of cystic fibrosis: randomised trial of stepwise v couple screening. BMJ 310:353-357.

Rowley PT, Loader S, Levenkron JC. 1997. Cystic fibrosis carrier population screening: A review. Genet Test 1:53-59.

Schwartz M, Brandt NJ, Skovby F. 1993. Screening for carriers of cystic fibrosis among pregnant women: a pilot study. Eur J Hum Genet 1:239-244.

Wald NJ, George LM, Wald NM, Mackenzie IZ. 1993. Couple screening for cystic fibrosis. Lancet 342:1307-1308.

Wald NJ, George L, Wald N, MacKenzi IZ. 1995. Further observations in connection with couple screening for cystic fibrosis. Prenat Diagn 15:589-590.

Witt DR, Schaefer C, Hallam P, Wi S, Blumberg B, Fishbach A, et al. 1996. Cystic fibrosis heterozygote screening in 5,161 pregnant women. Am J Hum Genet 58:823-835.