Chapter 6:

Chapter 6: Literature Search

A search of the medical literature was performed on a CD-ROM database (MEDLINE using PC-SPIRS) and articles discussing choroid plexus cysts and the ultrasound detection of abnormalities were reviewed. Further searching was done by cross-referencing from these articles, and by searching for specific chromosomal aberrations and topics which relate to genetic screening.

From this search 42 articles were obtained which pertain to fetuses with CPC or correlate CPC with chromosomal abnormalities from 1984 to 1994. These ranged from cases studies, to reviews of aneuploidy and ultrasound findings, to prospective studies with control cases. There was a total of 1,439 cysts in these articles from which much data could be obtained for analysis. Not all of these articles were suitable for each part of the analyses which follow in this thesis. For example, only articles which give details of cyst size were used to determine the utility of this criterion.

6.1 Association with Chromosomal Abnormalities

Without doubt the most frequent association was with T18; 129 cases were noted. There were 17 cases of T21 described, two cases of T13, three cases of triploidy, and 11 assorted autosomal or sex chromosomal aberrations. In this final group, five were del14+t(14q21q), in which the short arms of chromosome 14 were deleted and translocated to chromosome 21. This is in fact a rare type of Down syndrome said to occur in only 3 - 5% of Down syndrome cases. An explanation for the increased prevalence of this anomaly cannot be provided by this research. There were two cases of Klinefelters syndrome, one each of Turner syndrome, T9, a translocation of T13, and an unstated case. The types of aneuploidy are shown in Fig 10 and the rate of aneuploidy calculated in Table 3.

Fig 10: Types of aneuploidy noted in all fetuses with choroid plexus cysts.

Total	Aneuploidy	T18	T21	T13	Triploidy	Other
1693	Rate	7.62%	1.0%	0.12%	0.18%	0.95%

Table 3: Rates of aneuploidy in all fetuses with choroid plexus cysts.

The rate of T18 amongst fetuses with CPC is 7.6% in this meta-analysis, which seems an inordinately high figure. Similarly the 1% rate of T21 appears frightening. If these figures were true, the controversy would be over. The association with chromosomal abnormalities would be overwhelming and even obvious in clinical practice. However this is certainly not so, and the figures are skewed due to the high number of case studies and aneuploidy reviews in this group, with the addition of a publishing bias. Several items have to be factored in to the calculations if we are to have meaningful numbers to review.

6.2 Incidence of CPC

In order to obtain a better idea of the rate of aneuploidy, those articles which provided details of the size of the population were assessed. This effectively removed reviews of aneuploidy and case studies from the data set. There were 18 articles in which the incidence of CPC was given or could be calculated. In this set, there were 922 CPC from a population of approximately 92,879 patients, noted at an average incidence of 0.99% (0.86 - 3.5%). Of these, 45 (4.88%) were aneuploid. A summary of these figures is given in Table 4. The aneuploid rate can be equated to the statistical sensitivity.

Author	No. CPC	Incidence	Population	No. Aneuploid
Ostlere¹⁴⁴	11	0.30%	3627	0
Furness⁸¹	30	0.67%	4500	3
Chitkara⁴⁷	41	0.65%	6288	1
Clark⁵⁰	5	0.18%	2820	0
DeRoo⁶²	17	0.82%	2084	0
Gabrielli⁴³	82	2.30%	3565	4
Chan⁴⁵	13	2.50%	513	0
Camurri⁴⁰	10	0.33%	3000	1
Ostlere(90)¹⁴⁶	100	0.85%	11700	3
Chinn⁴⁶	38	3.60%	1045	1
Twining¹⁸⁹	19	0.42%	4541	2
Platt¹⁵	71	0.96%	7350	4
Achiron²	30	0.55%	5400	5
Perpignano¹⁵⁰	87	2.30%	3769	6
Howard	51	1.07%	4765	1
Porto¹⁵⁴	63	1.90%	3247	7
Kupfermine¹¹⁶	102	1.1%	9100	3
Walkinshaw¹⁹³	152	0.98%	15565	4
Total	922	0.99%	92879	45

Table 4: Articles in which incidence of CPC was given or could be determined.

The types of aneuploidy are shown in Fig 11 and the rates calculated in Table 5.

Fig 11: Types of aneuploidy noted in articles where the incidence of cysts could be determined.

Aneuploidy	T18	T21	T13	Triploidy	Other
No. (Rate)	35 (3.8%)	6 (0.65%)	0	1 (0.11%	3 (0.33%)

Table 5: Rate of aneuploidy in articles where incidence of cysts could be determined.

With consideration of the incidence of CPC , the rates of aneuploidy are considerably lower than that found in the complete analysis. However the rate still appears quite high. For example, in our 49 cases from Geelong we would have expected two new cases of T18 if 3.8% was in fact the aneuploidy rate for CPC. At this point we need to consider the prevalence of T18 in the pregnant population, and factor this into the calculation as well.

6.3 Rate of Aneuploidy versus Risk of Aneuploidy

By this method we can obtain not so much a rate of aneuploidy, but the actual risk which the presence of a CPC gives for the likelihood of T18. We cannot use the figures in Table 5 for the clinical day-to-day practice of sonography and assessment of risk. Are one in 25 cases of CPC (~4%) really going to be T18? Of course not.

According to the data supplied by the PDCU for the years 1991-1993† , the rate of occurrence of T18 in the Victorian population of fetuses is 1/2311 or 0.0433%. This figure corresponds with other estimations of prevalence of T18. Snijders¹⁷⁷ has determined charts for the background risk of various chromosomal abnormalities referred against maternal age and fetal gestational age. For a 29 year old woman at 16 weeks gestation (the usual age for amniocentesis) the risk is calculated to be 1/2371. These charts are reproduced in the Appendix. Gross⁹¹ also calculated the background risk of T18, basing her calculations on the birth data of Hsu (referenced in Gross’s article) and the 67.5% loss rate between amniocentesis and delivery reported by Hook at al (also referenced in Gross’s article), and found the T18 incidence to be 1/2461.

Based on these figures, from 922 low-risk patients we would ordinarily expect 0.39 cases of T18. Therefore there does appear to appears to be a very great increase in the rate of aneuploidy, in particular T18, in fetuses with CPC. But we if combine these sensitivity rates with the incidence of cysts as well as the background risk for T18, we can determine the likelihood ratio and use Bayes Theorem to determine the much more useful odds or risk of aneuploidy based on the normal population. This has the effect of overcoming much of the bias caused by different risks in the populations in the published articles and in the real world, and therefore by any publishing bias.

Odds (Risk) of T18 in a fetus with CPC:

prevalence x incidence / incidence of CPC

0.000433 x 0.038/ 0.0099 = 0.00166 or 1 in 601.

CPC increases the odds of T18 about 3.8 times.

The odds of T18 in fetus with a CPC are increased about 3.8 times beyond the background risk, but at 1/601, the risk is much lower than the risk for amniocentesis. The implications of amniocentesis are discussed in later chapters. This calculation provides a statistically valid method of analysing the data provided in the literature. Unfortunately most articles fail to complete this calculation, and therefore the implications of CPC are not fully explained to the readers.

6.5 Differences in the referral population and indications for scanning

Do the articles under analysis discuss a population with same risk as the one we find in our workplaces? In Australia there is no limitation to the indications for scanning pregnant women. This is reflected in the very high utilisation of ultrasound in pregnancy in Australia, approximately 97% of Victorian women²⁴⁶, and 7 of 10 women having a “routine scan” at 18 weeks of gestation, most without any special clinical indication. In the United States of America, pregnancy ultrasound is often limited to a more strict set of indications, which must be adhered to if insurance payment is to be guaranteed. Approximately 60 – 70% of pregnant women in America have an ultrasound scan⁸⁸. It could be reasoned therefore that data from the USA already represents a higher risk population as the patients have a clinical indication.

The rate of chromosomal malformations in Victoria was 0.43% for the year 1993⁵⁶ with 10% of these being T18. From the population in the literature of 92,789 patients, we would expect 398 cases of aneuploidy, of which 40 would be T18. As only 45 cases of aneuploidy were reported, of which 35 wereT18, this would indicate that the population under study has an extremely low rate of aneuploidy in general, but appropriate risk for T18. It could be reasoned that in this population, there is no increase in risk for aneuploidy. But unfortunately discussion must be limited because many of these articles only relate the aneuploidies in those fetuses with CPC. The rate of aneuploidy in the remainder of their patients is often unstated, and this would of course affect our comparisons of the background risk.

Only one of the studies used in this section (Chan⁴⁵) stated it had a primarily “high-risk” population in that 79% of their referral population was over 35 years of age. Many articles however came from tertiary referral centres, where a mix of high and low risk populations are seen. These are much different from the screening population as seen by most sonographers in Australia, and this is reflected in the high rate of amniocentesis in some studies.

6.6 Risk in Down Syndrome (T21)

Some authors, particularly Kupfermine¹¹⁶ and Zerres²⁰⁰, have suggested that as Down Syndrome has been reported in some fetuses with CPC, that is enough to warrant amniocentesis in all cases. The prevalence of T21 at 16 weeks in a 29 year old has been calculated by Snijders¹⁷⁷ to be 1/ 695, or 0.143%. The presence of 6 T21 fetuses in 922 cases of CPC (0.65%) would appear to suggest an increase risk of T21 by 4.5 times. However again we must temper this with the difference in referring protocols, the risk in this particular population and the prevalence of T21.

We can use Bayes Theorem to calculate the true risk of T21, given the incidence of CPC and of T21 in the normal pregnant population.

Odds (Risk) of T21 in a fetus with CPC:

prevalence x incidence / incidence of CPC

0.00143 x 0.0065 / 0.0099 = 0.00098 or 1 in 1018.

CPC decreases the risk of T21 by about half!

There are in fact fewer T21 cases in this population than we would expect, given the known incidence. It would appear that CPC actually protects against T21.

The ultrasound detection of T21, while relatively successful in some hands, represents the Holy Grail of screening sonography. Ultrasound has not been good overall in detecting T21 at the 18-20 wks scan due to the subtlety of the structural defects in T21 as opposed to the rather gross malformations seen in many cases of T18 and T13^{181, 51}.

The six cases of T21 can be explained by the high rate of amniocentesis in the relevant articles. These are shown in Table 6. With these exceptionally high amniocentesis rates, T21 is certain to be found either as a chance, incidental finding due to the background risk (which is unobtainable due to lack of information concerning maternal age, but we must to be high given the amniocentesis rate), and particularly when other malformations are present, as they were in three of the six cases.

	Amnios	Fetuses	Amnio Rate
Platt¹⁵²	62	71	87.3%
Perpignano¹⁵⁰	83	87	96.4%
Porto¹⁵⁴	56	63	88.9%
Nava1²⁵	176	211	83.4%
Kupfermine¹¹⁶	79	102	77.4%
Walkinshaw¹⁹³	54	152	35.5%

Table 6: Rates of Amniocentesis is series where T21 was detected.

The supposed relationship with Down syndrome is almost certainly an incidental and statistical one generated by these high rates of amniocentesis. The higher than expected proportion of fetuses with translocation T21 cannot be explained. However, analysis with Bayes Theorem in fact points to a protective effect of CPC against T21!

6.7 Conclusion

There certainly appears to be an association of CPC with T18, but not with other chromosomal abnormalities. The strength of the association does not appear to be sufficient to warrant amniocentesis in the younger age groups, but depending upon the maternal age this may be modified. The risk for a 29 year old woman is definitely lower than the loss rate for amniocentesis. But are there characteristics of the CPC which modify this risk? Is the paradigm appropriate?