The pathophysiology of human premature cervical remodeling resulting in spontaneous preterm birth: Where are we now?

With the introduction of cervical length measurement, it has become clinically evident that when cervices start to shorten in pregnancy, this process is usually characterized by funneling or dilation/failure of the internal os while the external os remains closed. ( 83 ) In addition, we know that after a vaginal delivery of a fetus, the internal os rapidly closes while the external os stays dilated (sometimes for weeks or months or even into subsequent pregnancies). Recently, our lab re-evaluated whole transverse slices of normal, nonpregnant cervical tissue from the internal os, mid-cervix and external os to further understand why these regional differences in cervical tissue function exist. ( 11 ) Unlike the prevailing paradigm which characterizes the human cervix as a mostly homogenous, collagenous structure with minimal cellular content, we found that there is an abundant amount of smooth muscle at the internal os (about 50–60% of the tissue) and that there are bundles of smooth muscle that circumferentially oriented around the periphery of the cervix. ( 11 ) It is interesting to note that this muscle structure mimics the collagen network “zone” that was reported by Yao et al. ( 14 ) This smooth muscle orientation persists as one moves down to the mid-cervix. The smooth muscle content then gradually decreases in the lower half of the cervix. At the external os, smooth muscle can be found scattered in the tissue and accounts for about 10% of the tissue content. ( Figure 4 , [11]) In addition, we found that whole slices of cervical tissue from the internal os are more contractile than tissue obtained from the external os in functional organ baths. ( 11 ) Recent rodent studies have also shown that the rodent cervix contains a significant amount of smooth muscle and that the cervix does contract in response to contractile agonists. ( 84 ) Given these findings, as well as the fact that the cervix remains highly innervated in pregnancy while uterine innervation decreases, and that the internal os is more densely innervated than the external os, it is intriguing to consider the possibility that the smooth muscle at the internal os may form a specialized sphincter. If the internal os is a sphincter, it is then reasonable to ask, “is cervical funneling or dilation of the internal os in pregnancy actually ‘sphincter failure’”? Along the same line, when the internal os closes rapidly after vaginal delivery, is the sphincter resuming is original shape? Going one step farther, some studies have shown cesarean sections done in the second stage of labor may increase the rate of sPTB in subsequent pregnancies. ( 85 ) If the sPTB in subsequent pregnancies are due premature cervical failure, could it be the “sphincter” was damaged or altered at time of cesarean section? Given the structure and clinical function of the internal os, these concepts make sense biologically but further studies are needed. The concept of a sphincter in the cervix is actually not new. Several decades ago, studies questioned why smooth muscle cell existed in the cervix. In the 1950s, a study showed in response to contractile agonists, the pregnant cervix is able to contract separate from the uterus.( 86 ) Several studies showed the cervix exhibits electromyography (EMG) activity and cervical EMG activity correlates to the length of the latent phase of labor. These studies suggested the cervix can contract and prolong the length of the latent phase of labor. ( 87 – 93 ) Unfortunately, over time (and likely due to the assumption that the cervix was a mostly homogenous, collagenous structure), evaluating cervical smooth muscle function fell out of favor as more attention was given to evaluating if cervical failure in pregnancy was due to defects in the ECM.

Approximately one in ten (approximately 500,000) pregnancies results in preterm birth (PTB) annually in the United States. ( 1 ) Since premature babies are at significant risk of neonatal demise, complicated and prolonged NICU stays, lifelong disabilities, and chronic illnesses, the problem of PTB not only presents an emotionally challenging time for families impacted by a preterm birth, but it is also an enormous financial burden on US Healthcare. Caughey et al recently reported that based on cost estimates in 2006, the cost of PTB in 2015 was approximately $31 billion. ( 2 ) Although we have seen a slight decrease in the US PTB rate between 2007 to 2014, data from 2014 to 2015 shows the preterm birth rate has slightly increased. ( 1 , 3 ) It is even more intriguing to note that the rate of PTB has not significantly decreased since the 1980s. ( 4 , 5 ) Our field of Obstetrics has not been able to make an impactful and lasting dent in the rate of PTB in part for the following reasons. First, the etiology of sPTB likely involves extremely complex interactions between numerous factors including but not limited to genetics, hormones, the immune system, reproductive tissue properties (uterus, cervix, fetal membranes, placenta), vascular system, nervous system, maternal anatomy (geometric factors), our microbiome and the environment. In order to understand the impact of all these factors on pregnancy, we must start to look at the problem of preterm birth from a multidisciplinary perspective. Second, definitions of preterm birth (i.e. gestational age cutoffs) and outcomes of interest have not been standardized in studies making generalizability and comparability difficult. Along these same lines, there are several “phenotypes” of sPTB (i.e. those that start with premature activation of the fetal membranes leading to premature membrane rupture vs premature activation of the myometrium leading to preterm labor vs premature cervical remodeling leading to cervical failure) and we cannot assume that these phenotypes should be lumped into one category when analyzing sPTB. ( 6 ) Third, because pregnancy is a “protected state”, progress in our field has been thwarted by the lack of ability to study gestational-age matched normal and abnormal human reproductive tissues (i.e. placenta, myometrium, fetal membranes, cervical tissue) at various time points in pregnancy. The ideal progression in science is first to understand normal tissue structure and physiology, then characterize how normal tissue structure and function goes awry in disease states. Once we are able to detect these changes, we can then identify targets to alter the mechanistic pathways to either prevent or revert pathology back to the normal state. Since it is challenging, if not impossible, to obtain and test normal and gestational age-matched abnormal tissues in pregnancy, our field is left to make certain assumptions regarding mechanistic pathways leading to sPTB. Could this be one reason why our current therapies are not always effective? Although the inciting factors leading to sPTB vary, the final pathway to sPTB inevitably must involve premature remodeling / softening / shortening of the cervix leading to dilation of the cervix and delivery of the fetus. This point is exemplified by the fact that if the fetal membranes were to prematurely rupture or the patient went into preterm labor, the fetus would not be delivered if the cervix remained closed. In order to decrease the rate of sPTB, it is imperative that we improve our understanding of normal and abnormal reproductive tissue structure and function and how these tissues interact with each other at a cellular and biochemical level. Since other chapters in this issue will be focusing on the myometrium and fetal membranes, the goal of this chapter is to focus on the compartment of the cervix. We will review the current literature on normal and abnormal human cervical tissue remodeling and identify gaps in knowledge. Our goal is also to introduce a revised paradigm of normal cervical tissue structure and function which will provide novel research opportunities that may ultimately lead to developing safe and effective interventions to significantly decrease the rate and complications of prematurity.

This review introduces a revised paradigm of cervical tissue structure and function that includes the possibility of a specialized sphincter in the cervix. This paradigm states the cervix is not a passive, homogenous, collagenous structure and resident smooth muscle cells in the stroma may play an active role in cervical function. This paradigm opens new doors for exploration which include but are not limited to the following. 1) What is the contractile role of cervical smooth muscle cells at the internal os in normal pregnancies and pregnancies complicated by premature cervical failure? Similarly, cervical smooth muscle cells are contiguous with uterine smooth muscle cells and likely have the ability to send signals to the uterus. As such, could they have a role in orchestrating or propagating uterine contractions? 2) We know that smooth muscle cells from other body systems (i.e. vascular and uterine smooth muscle cells) can produce ECM and enzymes involved in remodeling ECM. ( 94 – 97 ) If cervical smooth muscle cells share similar capabilities, what is their role in maintaining ECM homeostasis in the cervix? 3) As the fetus grows, the weight of the pregnancy stretches the cervix, particularly at the internal os. ( 98 , 99 ) Towards the end of a normal pregnancy, Braxton-Hicks and labor contractions occur which further stretch the cervix. Given that i) tissue stress and stretch are an important mechanical factors in pregnancy, ii) smooth muscle cells can produce ECM and enzymes involved in remodeling ECM ( 94 – 97 ) and iii) uterine smooth muscle stretch can influence myometrial tissue remodeling ( 95 – 97 , 100 ), could cervical smooth muscle stretch influence cervical remodeling? Further, if cervical smooth muscle cells have altered ‘stretch’ sensors, could this trigger premature remodeling? Lastly, areas that also have largely been forgotten but are in dire need of investigation are the role of the vascular and nervous system in normal and abnormal cervical tissue remodeling. As we embark on studies investigating these novel areas, it is imperative that we move forward with clear and standardized terminology. Since “cervical insufficiency” can be a confusing clinical term, recently it was proposed to adopt the term “premature cervical remodeling” as it does not rely on clinical history but focuses on the physical changes in the cervix. ( 6 ) In order to use this terminology, however, we must establish objective detection and quantification methods to diagnose premature cervical remodeling. Currently, several groups are working on developing non-invasive methods to quantify cervical remodeling and these are discussed in the second issue of this series in Seminars in Perinatology. It is also imperative that we analyze normal and abnormal human tissues in pregnancy and that our in vitro models mimic the pregnant state (i.e. the hormonal milieu of pregnancy). Lastly, but most importantly, it is essential that we remember that each “compartment” (i.e. uterus, cervix, fetal membranes, immune/vascular/nervous systems, environment etc.) does not operate in isolation. Future studies will need to take into account interactions between “compartments”. Moving forward, the path to success is one that involves active, multidisciplinary collaboration and communication between experts in various fields that include but are not limited to geneticists, epidemiologists, basic scientists, clinicians, systems biologists and engineers. With such an approach, we may finally identify novel and effective interventions to significantly decrease the rate and complications of prematurity.

The cervix is the mechanical barrier located at the bottom of the uterus that keeps the uterus closed during pregnancy. As a pregnancy grows, the cervix must withstand tissue stresses generated from a complex set of forces from the weight of the growing pregnancy and pull of the uterine wall. ( 7 ) Our understanding of human cervical tissue structure was established in 1940s when Danforth et al reported that the cervix is a predominantly homogenous, hydrated, collagenous structure (about 85–90% collagen/extracellular matrix [ECM] that contains proteoglycans, glycosaminoglycans, elastin, matricellular proteins) and a minimal amount of cells (about 10–15%) which includes immune cells, fibroblasts, smooth muscle cells, and glandular/vascular cells. ( 8 – 10 ) In addition, over time, an assumption has been made that there may be a distinct line that demarcates where the uterus (smooth muscle body) ends and the cervix (collagen/ECM) begins. ( Figure 1 ; [11]) Similar to other load-bearing tissues in the body (i.e. bone, tendon), the mechanical strength of cervix is thought to rely mainly in part on the collagen network in the ECM. Studies in the 1970s reported that the collagen network in the cervix is comprised of three distinct zones – one zone of collagen in the middle of the stroma that is oriented circumferentially around the endocervical canal (which provides resistance to cervical dilation) and an inner (toward the endocervical canal) and outer (towards the vagina) zone of collagen fibers that run parallel to the endocervical canal These out and inner zones of collagen likely attach the cervix to the uterus. ( 6 , 12 , 13 ; Figure 2 ) Recently, Yao et al used optical coherence tomography (OCT) to measure the orientation of collagen fibers and their distribution in the upper cervix using hydrated unfixed samples, and the study reported in general there are two radial zones and the outer zone contains collagen fibers that are circumferentially oriented. ( 14 ; Figure 3 ) Reusch et al used second harmonic generation microscopy to show that collagen fibers in the cervix form interweaving zones that change as one moves from the external to internal os. ( 15 ) In addition to the overall structure of the collagen network, the strength of a collagen network also depends on collagen fiber alignment and the degree and type of cross-linking that is present between collagen fibrils.( 16 , 17 ) Recent work by Zork et al has shown that collagen crosslinks can be measured in the human cervix and that the internal os exhibits more collagen crosslink heterogeneity compared to the external os.( 18 ) Both the work by Reusch et al and Zork et al stress the point that unlike the prevailing paradigm, the cervix is not a homogeneous structure and that regional differences in tissue composition do exist between the internal and external os. Other important components of ECM that influence its overall mechanical strength, include proteoglycans/glycosaminoglycans and matricellular proteins. Proteoglycans (ie decorin, versican, fibromodulin, biglycan, asporin) are important for organizing collagen fibrils and glycosaminoglycans (ie hyaluronic acid) influence tissue hydration and thus tissue compliance. ( 16 , 19 – 25 ) Matricellular proteins (i.e. tenascins, thrombospondins, SPARC proteins) are rapidly turned over during the remodeling process and can regulate matrix formation and cell-matrix interactions. ( 16 , 26 ) In addition to the stromal ECM, normal cervical tissue architecture includes the endocervical canal, which is lined by a single layer of mucin-secreting columnar epithelium, which form deep clefts or mucinous glands. The ectocervix (vaginal portion of the cervix) is covered with nonkeratinized stratified squamous epithelium. The junction between the endo- and ecto-cervical cells is the squamocolumnar junction. ( 27 ) The endo- and ectocervical cells are thought to play an important barrier function to protect against ascending infection from the vagina. ( 28 , 29 ) Although most of the previous work evaluating cervical tissue structure has focused on the ECM, we also know the cervix is vascular and blood flow to the cervix provides a means of oxygenation and communication (i.e. via hormones, immune cells, chemokines/cytokines etc) from the rest of the body. The blood supply to the cervix is provided by branches of the uterine artery. A recent study by Bereza et al showed that in the nonpregnant state, there are four distinct “zones” of vessels in the vaginal and supravaginal apsects of the cervix. There is an outer zone which contains large arteries and veins, a middle zone (in the stroma) that contains arterioles and venules, an endocervical mucosal capillary zone and a zone parallel to the endocervical canal that contains small capillaries and veins. ( 30 – 34 ) How or if the cervical vasculature changes throughout pregnancy remains unknown. Even less is known about cervical tissue lymphatic changes in pregnancy. Another interesting and important avenue of communication from the rest of the body to the cervix is via the nervous system. Although it is unclear why this is the case, the cervix is one of the most densely innervated structures of all the reproductive organs. Innervation to the pelvic organs is predominantly autonomic with sympathetic innervation provided by the superior hypogastric plexus (presacral nerve) which originates from spinal levels T 10 -L 2 . Parasympathetic innervation is derived from neurons originating from spinal levels S 2 – S 4 and axons from these spinal nerves combine to form the pelvic splanchnic nerves. The hypogastric (sympathetic) and pelvic splanchnic (parasympathetic) nerves combine to form the inferior hypogastric plexus (or the pelvic plexus) which further divides into branches that innervate the various pelvic organs including the cervix. The cervix also contains afferent sensory nerve fibers which transmit painful stimuli through the pelvic splanchnic nerves to the sacral nerves (S 2–4 ). ( 35 ) Although this is still an emerging area of research that is ripe for further investigation, a few studies have shown that the human cervix is more densely innervated than the uterus and in pregnancy that uterine innervation decreases significantly 30-fold compared to non-pregnant uterine tissue) while the degree of innervation in the cervix remains the same. ( 36 – 42 )

To date, studies that have evaluated the mechanisms leading to cervical “failure” in pregnancy have been inconclusive in part because diagnostic criteria to define a “weak” cervix remain elusive. Traditionally, we have used the clinical term “cervical insufficiency” or “an incompetent cervix” to define a condition where the cervix tends to asymptomatically dilate repeatedly in the mid-trimester leading to a spontaneous abortion or periviable/premature delivery. ( 73 ) Due to recent studies that have shown that premature cervical shortening significantly increases a woman’s risk for sPTB, many now consider premature cervical shortening as a sign of “cervical insufficiency”. ( 74 ) Although studies have identified risk factors for “cervical insufficiency” ( 73 ), studies that have obtained and evaluated human cervical tissue to determine the underlying biochemical cause of “weak cervical tissue” in pregnancy remain inconclusive. Table 1 lists the studies that have evaluated human cervical tissue in women with “cervical insufficiency” and sheds light on several issues. ( 75 – 82 ) First, there is a paucity of data evaluating human cervical tissue in women with cervical insufficiency. Second, some of these studies defined “cervical insufficiency” differently and methods for identifying those with “cervical insufficiency” in some studies may not be ideal or accurate (i.e. by ICD-10 codes). Third, cervical biopsies were not obtained in the same location. As studies are beginning to show, the cervix is not a homogenous structure and biopsy location does matter. Fourth, timing of cervical biopsy in pregnancy were different. In most studies, cervical biopsies were obtained in the nonpregnant state or postpartum after the cervix has undergone cervical dilation and delivery of the fetus. Some studies obtained biopsies during pregnancy in patients with “cervical insufficiency” but the control patients were either nonpregnant women undergoing benign hysterectomy or not gestational-age matched controls in pregnancy. These reasons might influence why we have not been able to find conclusive results regarding if there is an ECM defect in cervical tissue from women who exhibit premature cervical failure.

For most of pregnancy, the cervix must remain strong and closed to keep the fetus in utero. Then in a dramatic reversal of roles, at the end of pregnancy, the cervix must be compliant enough to dilate and allow for delivery of the fetus. Once the fetus is delivered, the cervix goes from being fully dilated to having a closed internal os – a process than can occur within minutes. Notably, while the internal os closes, the external os usually remains loose and open. In subsequent pregnancies, the external os may continue to remain soft and dilated while the internal os is firmly closed. This common finding is usually described by clinicians as a “multiparous cervix”. Clinically, we can feel the cervix start to soften (Hegar sign) and see increased cyanosis as early as one month after conception. ( 43 ) Studies evaluating the mechanical properties of human cervical tissue in the non-pregnant, pregnant and postpartum state have confirmed our clinical findings that cervical tissue does indeed soften as pregnancy progresses. ( 44 – 47 ). Unfortunately, studies that evaluate how the human cervix changes at a biochemical level throughout human pregnancy is lacking as it is challenging to obtain cervical tissue (whole cervix tissue or biopsies) at various time points throughout pregnancy. Thus, the data that does exist regarding how the cervix remodels in pregnancy at a biochemical level have resulted from rodent studies. This review provides a brief overview of the phases of cervical remodeling that has been shown to occur in rodents. For comprehensive details, please refer to the excellent reviews by Word et al ( 25 ), House et al ( 44 ), Nallasamy et al ( 48 ), Mahendroo et al ( 49 ) and Timmons et al ( 50 ). In short, cervical remodeling in rodents has been characterized to consist of four overlapping phases. The first phase is “cervical softening” which is described as a prolonged, progressive softening phase that starts early in pregnancy. This is a finding that we see clinically with the Hegar sign that starts within the first month of pregnancy. Although the cervix softens, the cervix stays closed because the mechanical forces acting on it are not large enough to dilate it. This delicate equilibrium balance most likely differs for every patient depending on cervical stroma consistency, the weight of the fetus, and maternal anatomy. ( 47 ) The cervical softening phase in rodents has been shown to involve decreases in collagen solubility (which is an indirect method to evaluate collagen crosslinking) and a shift to include more immature (weaker) collagen crosslinks vs mature (stronger) collagen crosslinks. ( 16 , 17 , 51 ) Yoshida et al reported that these alterations in the ratio of mature to immature collagen crosslinks directly relates to changes in cervical tissue compliance. ( 17 ) Akins et al reported that the cervical softening phase also includes decreases in matricellular proteins, thrombospondin 2 and tenascin-C. ( 16 ) As rodent pregnancies approach term, the cervix undergoes a more rapid “ripening” phase, which results in increased cervical tissue compliance or weakening to prepare for cervical dilation. ( 25 , 44 , 49 , 50 ) At a biochemical level, this phase is characterized by changes in the organization, shape and size of collagen fibers and an increase in matrix metalloproteinases (MMPs) which are enzymes that degrade ECM. ( 52 , 53 ) There is also an increase in the amount of hyaluronic acid, which is thought to increase tissue hydration and may be involved in disrupting or weakening of the collagen network. ( 25 , 29 , 49 – 51 , 54 ) However, preliminary studies by Akgul et al found that HA is not essential to normal cervical ripening. Due to the fact that HA has such a dramatic shift in content nearing term, the biomechanical role of HA still needs to be elucidated. ( 29 ) Several studies have suggested that normal cervical ripening may be a sterile inflammatory state characterized by an influx of immune cells into the cervix. ( 55 – 65 ) However, although studies have shown an influx of immune cells into cervical tissue, it is unclear what triggers the immune cells to travel to the cervix. In addition, evidence from other studies suggest that immune cell activation is not vital for cervical ripening. ( 66 – 69 ). Interestingly, several studies have investigated the role of the nervous system in cervical remodeling and parturition and these studies propose that nerves in the cervix may play a role in inducing a “sterile” neurogenic inflammatory state. ( 70 – 72 ) Given that the cervix remains densely innervated throughout pregnancy, this is a tantalizing hypothesis that is in dire need of further investigation. Once the cervix is “ripe” it can now dilate as the uterus goes into labor and allow for delivery of the fetus. The last phase of cervical remodeling is “postpartum repair”. ( 25 , 49 ) To date, detailed data on how the cervix dramatically goes from a fully dilated structure to have a closed internal os in a matter of minutes is lacking. Although extensive and extremely detailed work has evaluated the phases of cervical remodeling in rodents, we must keep in mind that rodents do not share the same anatomical structure of their reproductive organs with humans and rodents are quadrupedal mammals which undergo different gravitational forces on their pregnant pelvic organs compared to bipedal humans that stand upright. In addition, hormonal (estrogen and progesterone) changes that are seen in rodents also do not perfectly mimic changes seen in humans. Regardless, since obtaining cervical tissue during human pregnancy is challenging, insight obtained from rodent studies are vital to establishing a guide as to what may possibly be occurring in human cervical remodeling.