Timothy J Hall, Helen Feltovich, Lindsey C. Carlson, Quinton W. Guerrero, Ivan M. Rosado-Mendez
University of Wisconsin, Madison, WI (tjhall@wisc.edu)
Click here to read the abstract

The cervix, which is the opening into the uterus, is a remarkable organ. It is the only structure in the entire body that has diametrically opposed functions. The normal cervix stays stiff and closed throughout most of pregnancy while the baby develops and then completely softens and opens at just the right time so the full-grown baby can be born normally, through the vagina. This process of softening and opening involves remodeling, or breaking down, of the cervix’s collagen structure together with an increase in water, or hydration, of the cervix. When the process happens too soon, babies can be born prematurely, which can cause serious lifelong disability and even death. When it happens too late, mothers may need a cesarean delivery. [1,2]

Amazingly, even after 100 years of research, nobody understands how this happens. Specifically, it is unclear how the cervix knows when to soften, or what molecular signals initiate and control this process. As a result, obstetrical providers today evaluate cervical softness in exactly the same way as they did in the 1800s: they feel the cervix with their fingers and classify it as ‘soft’, ‘medium’ or ‘firm’. And, unsurprisingly, despite significant research effort, the preterm birth rate remains unacceptably high and more than 95% of preterm births are intractable to any available therapies.

The cervix is like a cylinder with a canal in the middle that goes up into the uterus. This is where the sperm travels to fertilize the egg, which then nests in the uterus. Cervical collagen is roughly arranged in 3 layers. There is a thin layer on the inner part that where collagen is mostly aligned along the cervical canal, and similarly a thin layer that runs along the outside of the cervix. Between these layers, in the middle of the cervix is a thicker layer of collagen that is circumferential, aligned like a belt around the canal. Researchers theorize that the inner and outer layers stabilize the cervix from tearing off as it shortens and dilates for childbirth, while the middle circumferential layer is most important for the strength and stiffness of the cervix to prevent it from opening prematurely, but nobody is certain. This information is critical to understanding the process of normal, and abnormal, cervical remodeling, and that is why recently there is growing interest in developing technology that can non-invasively and thoroughly assess the collagen in the cervix and measure its softness and hydration status. The most promising approaches use regular clinical ultrasound systems that have special processing strategies to evaluate these tissue properties.

One approach to assessing tissue softness uses ultrasound pressure to induce a shear wave, which moves outward like rings in water after a stone is dropped. The speed of the shear wave provides a measurement of tissue stiffness because these waves travel slower in softer tissue. This method is used clinically in simple tissues without layers of collagen, such as the liver for staging fibrosis. Cervical tissue is very complex because of the collagen layers, but, after many years of research, we finally have adapted this method for the cervix. The first thing we did was compare shear wave speeds in hysterectomy specimens (surgically removed uterus and cervix) between two groups of women. The first group was given a medicine that softens the cervix in preparation for induction of labor, a process called “ripening”. The second group was not treated. We found that the shear wave speeds were slower in the women who had the medicine, indicating that our method could detect the cervices that had been softened with the medicine.[3] We also found that the cervix was even more complex than we’d thought, that shear wave speeds are different in different parts of the cervix. Fortunately, we learned we could easily control for that by measuring the same place on every woman’s cervix. We confirmed that our measurements were accurate by performing second harmonic generation microscopy on the cervix samples in which we measured shear wave speeds. This is a sophisticated way of looking at the tiny collagen structure of a tissue. It told us that indeed, when the collagen structure was more organized, the shear wave speeds were faster, indicating stiffer tissue, and when the collagen started breaking down, the shear wave speeds were slower, indicating softer tissue.

The next step was to see if our methods worked outside of the laboratory. So we studied pregnant women who were undergoing cervical ripening in preparation for induction of labor. We took measurements from each woman before they got the ripening medicine, and then afterwards. We found that the shear wave speeds were slower after the cervix had been ripened, indicating the tissue had softened.[4] This told us that we could obtain useful information about the cervix in pregnant women.

We also evaluated attenuation, which is loss of ultrasound signal as the wave propagates, because attenuation should increase as tissue hydration increases. This is a very complicated measurement, but we found that if we carefully control the angle of the ultrasound beams relative to the underlying cervical structure, attenuation estimates are sensitive to the difference between ripened and unripened tissue in hysterectomy specimens. This suggests the potential for an additional parameter to quantify and monitor the status of the cervix during pregnancy, and we are currently analyzing attenuation in the pregnant women before and after they received the ripening medicine.

This technology is exciting because it could change clinical practice. On a very basic level, obstetrical providers would be able to talk in objective terms about a woman’s cervix instead of the subject “soft, medium, or firm” designation they currently use, which would improve provider communication and thus patient care. More importantly, it could provide a means to thoroughly study normal and abnormal cervical remodeling, and associate structural changes in the cervix with molecular changes, which is the only way to discover new interventions for preterm birth.

References
1. Feltovich H, Hall TJ, and Berghella V. Beyond cervical length: Emerging technologies for assessing the pregnant cervix. Am J Obst & Gyn, 207(5): 345-354, 2012.
2. Feltovich, H and Hall TJ. Quantitative Imaging of the Cervix: Setting the Bar. Ultrasound Obstet Gynecol. 42(2): 121-128, 2013.
3. Carlson LC, Feltovich H, Palmeri ML, Dahl JJ, Del Rio AM, Hall TJ, Estimation of shear wave speed in the human uterine cervix. Ultrasound Obstet Gynecol. 43( 4): 452–458, 2014.
4. Carlson LC, Romero ST, Palmeri ML, Munoz del Rio A, Esplin SM, Rotemberg VM, Hall TJ, Feltovich H. Changes in shear wave speed pre and post induction of labor: A feasibility study. Ultrasound Obstet Gynecol.published online ahead of print Sep.2014.

Franklin C. Lee¹ – franklee@uw.edu
Jonathan D. Harper¹ – jdharper@uw.edu
Thomas S. Lendvay^1,2 – Thomas.lendvay@seattlechildrens.org
Ziyue Liu³ – ziliu@iupui.edu
Barbrina Dunmire⁴ – mrbean@uw.edu
Manjiri Dighe⁵ – dighe@uw.edu
Michael Bailey⁴ – bailey@apl.washington.edu
Mathew D. Sorensen^1,6 – mathews@uw.edu

University of Washington
¹ Department of Urology, Box 356510
⁵ Department of Radiology, Box 357115
1959 NE Pacific St, Seattle, WA 98195

² Seattle Children’s Hospital
Urology, Developmental Pediatrics
4800 Sand Point WA NE, Seattle, WA 98105

³ Indiana University
Department of Biostatistics
410 W. Tenth St, Suite 3000, Indianapolis, IN 46202

⁴ University of Washington
Applied Physics Lab – Center for Industrial and Medical Ultrasound2
1013 NE 40th St, Seattle, WA 98105

⁶ Department of Veteran Affairs Medical Center
Division of Urology
1660 South Columbian Way, Seattle, WA 98108

Stone disease in the children is becoming more commonplace. Over the past 25 years the incidence has increased approximately 6-10% annually and is now 50 per 100,000 adolescents¹. The diagnosis of kidney stones in children, as in adults, relies primarily on diagnostic imaging. In adults, the most common imaging study performed in the United States for the initial diagnosis of kidney stones is a computed tomography (CT) scan, due to its superior sensitivity and specificity. This is not preferred for children as CT utilizes ionizing radiation and children have an increased sensitivity to radiation effects. In addition, stone formers often have recurrent stone episodes over their lifetime, which is especially relevant to younger stone formers. The repeated exposures of a CT scan could lead to an increased risks of secondary cancers². As a result, ultrasound is often performed in children instead because there is no radiation associated with its use¹. Ultrasound, however, is less sensitive and specific compared with CT, and is known to overestimate kidney stone size^3-5, which is one of the primary determinants of how stones are managed.

We have identified a new technique to improve the accuracy of US stone sizing. Traditionally, a kidney stone will show up as a bright, or hyperechoic, object on US, and radiologists measure the longest linear dimension of the bright area to represent the stone size. We believe that the dark, or hypoechoic, acoustic shadow that appears behind a kidney stone provides additional information for predicting stone size. The resolution of the stone is affected by the distortion of the waves traveling through the intervening tissues; the resolution of the shadow is only affected by the local stone obstruction.

We screened 660 stone diagnoses over an 11 year period (2004 – 2014) at Seattle Children’s Hospital, a tertiary care referral center serving the greater Pacific Northwest. Over the study period, there were 37 patients presenting with an initial diagnosis of a kidney stone, and who had both a US and CT within three months of each other. Two reviewers retrospectively measured both the stone size and the shadow width from the ultrasound image. We compared the results to the stone size measured from the CT scan. A total of 48 stones were included in the study with an average size based on CT imaging of 7.85 mm. The shadow width was present in 88% of the cases, and, on average, was more accurate than measuring the stone itself. Measuring the stone width on ultrasound tended to overestimate the size of the stone by 1.2 ± 2.5 mm (reviewer 1) and 2.0 ± 1.7 mm (reviewer 2), while measuring the shadow width on ultrasound underestimated the size of the stone by -0.6 ± 2.5 mm (reviewer 1) and overestimated the stone size by 0.3 ± 1.2 mm (reviewer 2). In both cases, the shadow was a better predictor of stone size and, for reviewer 2, there was less variability in the data. Stone sizes based on CT are typically considered within 1 mm, with low variability.

Our technique is simple and can be easily adopted by pediatricians, radiologists, and urologists. It improves the accuracy of US and gives physicians more confidence that the reported size is more representative of the true stone size, without having to expose children to the radiation of a CT scan. This also allows the physician to make more accurate decisions about when to perform a surgery for a large stone or continue to observe a small stone that may pass on its own. The findings from this study may also potentially be applicable to stones in adult patients. We believe that with continued advancements in US, we can reduce the number of CT scans for both adults and children. The results in part highlight one of the drawbacks of ultrasound, which is user dependence; two users can have very different results. It is anticipated that future work on automated stone and shadow sizing can reduce this along with standardizing the method in which the shadow measurement is taken.

References
1. Tasian G and Copelovitch L. Evaluation and Management of Kidney Stones in Children. J Urol. 2014: Epub ahead of print
2. Pearce MS, Salotti JA, Little MP, et al. Radiation exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours: a retrospective cohort study. Lancet 2012; 380: 499-505.
3. Ray AA, Ghiculete D, Pace KT, Honey RJD. Limitations to ultrasound in the detection and measurement of urinary tract calculi. Urology 2010; 76(2):295-300.
4. Fowler KAB, Locken JA, Duchesne JH, Williamson MR. US for detecting renal calculi with nonehnanced CT as a reference standard. Radiology 2002; 222(1):109-113.
5. Dunmire B, Lee FC, Hsi RS, Cunitz BW, Paun M, Bailey MR, Sorensen MD, Harper JD. Tools to improve the accuracy of kidney stone sizing with ultrasound. Aug 2014; [Epub ahead of print].