Preterm delivery especially before 34 weeks gestation is a major cause of morbidity and mortality of newborns worldwide, particularly in developing countries [2, 4, 6, 7, 39, 40]. The recently reported incidence of preterm in Thailand was about 10-13% . In our study, we had sPTD incidence of 8.5%, while other studies reported incidences between 7.8-10.6 % [42, 43]. There are numerous risk factors involved with preterm delivery such as a previous history of preterm birth, some medical diseases or hypertension associated with pregnancy[40, 42, 44, 45]. However, a considerable number of preterm deliveries have no clear risk factors, especially in the sPTD group [43, 46, 47]. Concerning the prediction of sPTD, many strategies have been proposed for use with basic clinical history [43, 46]. Cervical length measurement has been supported from many prior studies [46, 48-50]. However, it is still limited in some centers due to lack of specialists and being inconvenient to use as routine screening. Many laboratory investigations including maternal serum biomarkers, urine, and cervical mucus have been used to screening for sPTD [51-55]. However, to date no biomarker has been identified which can reliably predict sPTD as a general screening tool [52, 53, 56]. In Thailand, the quadruple test, offered by the National Health Policy free of charge, has been used for many years as a universal screening method for Down syndrome. Many studies have reported the extra benefit of the serum quadruple test as a predictor of adverse pregnancy outcomes, including preterm delivery [20, 21, 24, 25, 29, 36, 38]. However, the results of the serum quadruple test vary for the cutoff of each serum biomarker and it has to date shown limited use in predicting sPTD [20, 21, 24, 25, 29, 36]. Moreover, all of the prior studies assessing the use of serum biomarkers for prediction of preterm delivery included indicated preterm deliveries [20, 21, 24, 29, 36]. In our study, we wanted to clarify the association of quadruple biomarkers with sPTD, which has never been done in other studies.
When all serum quadruple biomarkers were combined to predict sPTD, predictive ability was low with an AUC of 0.56%, sensitivity of 55.6 % and false positive rate of 5% in total preterm deliveries. For early preterm deliveries, the sensitivity was 63.5 % with a false positive rate of 5%. However, a similar low percentage was found in the study in Thailand with triple serum biomarkers to predict preterm delivery as well . But the data of previous study included indicated preterm delivery . Our study found a relationship between high levels of AFP and relatively low levels of uE3 in early sPTD before 34 weeks, similar to the studies of Nunthapiwat et al  and preterm deliveries before 37 weeks in previous studies [20, 24, 25, 36]. In our study, the median serum level of AFP in the early sPTD group was 1.2 MoM, which was lower than in other studies which have reported levels between 1.34 and 2.0 MoM [20, 21, 25, 29, 36]. However, maternal serum AFP is mainly produced in the fetal yolk sac and liver, and high levels of this chemical have been associated with many fetal structural abnormalities and placental abnormalities  which were reported in the other studies [57-59]. The serum uE3 level was slightly lower in early sPTD pregnancies in our study, which has also been reported in other studies, varying between 0.5 and 0.9 MoM [20, 21, 24, 25, 29, 36]. The average serum level of uE3 in the early preterm group of our study was close to the study in a Thai ethnic group of Nanthapiwat et al. . Abnormally low levels of uE3 have also been reported as associated with fetal structural and placental abnormalities . These factors could be associated with indicated preterm and sPTD deliveries; however, we excluded pregnancies with fetal anomalies that can cause preterm delivery in our study.
When earlier models to predict sPTD using serum biomarkers were evaluated, they generally indicated that the association of maternal characteristics with serum biomarkers increased the ability to predict preterm delivery [21, 36]. Previous preterm delivery had a strong association with preterm delivery in many previous studies, which also recommended the administration of progesterone to reduce the risk of preterm delivery [10, 60-63]. Thus, combining a history of previous preterm with serum biomarkers can increase the rate of prediction of sPTD, as found in our study. We also found an association between sPTD and a short cervical length less than 35 mm. One recent study reported an association between preterm delivery and cervical length less than 25 mm before 24 weeks gestation . Our study proposes that the optimal model to predict sPTD combines serum AFP > 1.2 MoM and uE3 < 0.9 MoM with cervical length measurement less than 35 mm. and history of previous preterm delivery, which gave the highest ability to predict sPTD with an AUC of 0.86, sensitivity of 85% and a false positive rate of 10%. Our study suggests the quadruple test can also be uses as a national screening test for sPTD. When pregnant women are found to have an abnormally high level of AFP and a low level of uE3 whether or not a history of previous preterm delivery is found, these pregnant women should also have a cervical length measurement. Then, if the cervical length is less than 35 mm, progesterone should be administered to reduce the risk of sPTD. To select only women in the sPTD high risk group can reduce number of pregnant women who need to do a cervical length measurement, which is especially important for hospitals that lack a trained gynecologist who can measure cervical length as a universal screening tool for preterm delivery or in developing countries, where such women need to be referred to a larger center that has a specialist. After that, these high-risk pregnancy women should be monitored as a high risk for early sPTD in a tertiary center. The use of this proposed model will decrease the cost for measuring cervical length as a routine antenatal screening test without neglecting pregnancies at high risk of early sPTD, for which it is difficult to find an effective screening method. These are the extra benefits of serum biomarkers for Down syndrome screening which have never been reported before.
The remarkable strength of our study was that we included only sPTD patients while earlier studies included indicated preterm deliveries. In the indicated preterm delivery group, most of the causes can be identified from medical history or obstetrics complications. These conditions are associated with the indications for preterm delivery and can be predicted beforehand, and early management can be provided for prevention of sPTD. On the other hand, sPTD may involve many factors but a definite cause is difficult to clearly identify. Moreover, our study had only a small number of lost to-follow-up cases compared with most other studies. In addition, our study was conducted in the same setting as the laboratory tests were done, reducing the potential problem of serum measurement errors caused by offsite analysis. The current study originated from a prospective cohort that had retrospective analysis, thus the results were naturally blinded. Finally, our study was done in a homogeneous Thai population unlike other studies which have had mixed ethnicities. The cut off level of serum biomarkers we determined is appropriate for use as a universal screening for sPTD in Thailand without the need to adjust the serum levels according to various ethnicities. The notable limitation of our study was then small number of spontaneous early preterm deliveries, thus determining clear cut off levels for the serum biomarkers in this group needs further evaluation.