THE USE OF THREE-DIMENSIONAL ULTRASOUND IN GYNAECOLOGY
Three-dimensional (3D) ultrasound’s impact in gynaecology has been very gradual unlike in obstetrics where it is a fast growing modality. This is due to the added expense of introducing a new technology when two-dimensional (2D) sonography is already a valid diagnostic tool, and also because the benefits of 3D ultrasound in gynaecology are still uncertain. 3D ultrasound is capable of showing the clinician a multiplanar view, producing volume datasets which can be stored online for assessment at a later point in time. This allows display surface rendering of a region that provides a realistic image of a structure. 3D colour and power doppler sonography also offers the opportunity to show ovarian, endometrial, subendometrial and uterine vascularity. The main clinical applications are for the evaluation of the endometrium, ovaries, uterus and fallopian tubes. It is speculated to be able to distinguish between benign and malignant tumours in the endometrium and ovaries, introduce a new method for hysterosalpingo-contrast-sonography that is less time consuming and painful for the patient, and assist in fertility treatments to try and determine tubal patency and endometrial receptivity for in-vitro fertilisation (IVF) patients. This review looks at these main areas and demonstrates its development and usefulness within this field of ultrasound. It also critiques the literature currently available on the subject in an unbiased fashion to assess whether 3D ultrasound is of any diagnostic value within the gynaecological discipline.
Introduction and Background
1.1 PRINCIPLES OF ULTRASOUND
Ultrasound was first discovered to have a therapeutic effect in the 1950’s but only in the last three decades has it been used consistently in diagnostic medicine. (Meire & Farrant, 1982). It is an imaging technique that does not use ionising radiation and produces instant images, making it more suitable for certain procedures and patients. (Ernst & Feller-Kopman, 2006). The frequency of ultrasound is higher than what is audible by a human. Medical ultrasound has a frequency of 2-20 MHz, whilst a human can only generally hear sounds below 20 kHz. (Kurjak & Chervenak, 2006). An electric current is applied to tiny crystals in the transducer (probe), which causes it to vibrate and send out ultrasound waves. This is called the piezoelectric effect. A short pulse is sent into the body from the transducer, which waits for an echo to come back once the sound has hit a boundary between tissues. These echoes are then fed into the ultrasound machine which uses calculations from the time it took the echoes to come back and the speed of ultrasound in soft tissue (1540 m/s) to assign a grey scale level. These grey scale levels translate into a diagnostic image. (Meire & Farrant, 1982). The most used two-dimensional display mode is the B (brightness) mode and this produces information about the morphology of the area. Doppler ultrasound looks at the blood flowing through a particular area and has become very useful in gynaecology especially when measuring vascularisation. (Kurjak & Chervenak, 2006). Ultrasound has many uses, although it is most commonly known for its use in obstetrics. Sonograms are requested as the first line of investigation for various clinical problems including thyroid nodules, hepatic and renal masses, pregnancy screening, suspected prostate and testicular cancer and obviously certain gynaecological queries. (RCOR, 2003). This dissertation looks at the development of three-dimensional gynaecology and its uses today.
1.2 THREE DIMENSIONAL ULTRASOUND IN GYNAECOLOGY
“With three-dimensional ultrasonography, any desired plane through a pelvic organ can be obtained, regardless of the orientation of the sound beam during acquisition.” (Bega, Lev-Toaff, O’Kane, Becker & Kurtz, 2007, p.1249). This means that real-time surface rendering can be done to show an organ or volume and images can also be stored on computers so that they can be retrieved at a later date to by someone else other than the original sonographer. (Wu, Pan & Chang, 2007). The two-dimensional (2D) slices are put together to form three-dimensional (3D) volumes which are stored in units called voxels. (Smith, Chudleigh & Maxwell, 2005). Three-dimensional sonography can be displayed in the conventional grey-scale to demonstrate a volume or by colour or power doppler to estimate vascularisation in an area. (Raine-Fenning & Fleischer, 2005). Volumes are normally calculated with the aid of computerised formulas but as 3D ultrasound produces much more information, the result is more accurate. There are two methods used to calculate a volume – contour and the rotational method using Virtual Organ Computer Aided Analysis (VOCAL) software. (Raine-Fenning & Fleischer, 2005). The volumes of the uterus, cervix, ovaries, endometrium and subendometrium can all be generated to produce a 3D image on the display screen. Three-dimensional ultrasound is useful for imaging the uterus in the coronal plane which would shows the position of the uterus and any uterine abnormalities that may not be normally seen on two-dimensional sonography. (Timor-Tritsch & Monteagudo, 2007). Rendering of the uterus can also be done when “saline infusion sonohysterography (SIS) fills the cavity enabling the use of inversion rendering which creates a cast effect of the cavity.” (Timor-Tritsch & Monteagudo, 2007, p.167). This method allows for volumes to be calculated and stored and for other variables to be measured like endometrial thickness and polyp and fibroid, size and location. (Timor-Tritsch & Monteagudo, 2007). A similar method is used for assessing fallopian tube patency, where a contrast is injected into the uterus and observed travelling through to the fallopian tubes. (Exacoustos, et al., 2009). The interest in 3D ultrasound for successful vitro fertilisation grows by the day with the prospect of increasing pregnancy rates. (Järvelä et al., 2005). This has meant many studies in ovarian and endometrial blood flows and volumes. Blood flow through the ovaries, endometrium and sub-endometrium can be visualised with the introduction of 3D power Doppler which has enabled individual blood vessels to be measured and analysed using the VOCAL software which the produces blood flow indices. (Pan, Wu, Cheng, Li & Chang, 2002). These indices are vascularisation index (VI) which is the degree of vascularity; volume flow rate (FI) the signal intensity; vascularisation flow index (VFI) an amalgamation of VI and FI; and mean greyness (MG) which measures the echogenicity of the tissue. (Lam, Johnson & Raine-Fenning, 2007). These volumes and blood flows can also be used for the evaluation of adnexal masses and assessing whether a correlation exists with malignancy. (Alcazar, 2005). The uses are vast and extensive and whether any of these will become the ‘gold standard’ for any of the gynaecological procedures is still to be confirmed but it is definitely a useful and growing modality in a world where the desire to know more expands each day.
1.3 COMPARISONS IN TWO-DIMENSIONAL AND THREE-DIMENSIONAL ULTRASOUND IN GYNAECOLOGY
Two-dimensional ultrasound allows the visualisation of the transverse and sagittal planes of an area on the display, with the coronal plane only being visible rarely. (Bega et al., 2003). Three-dimensional ultrasound allows the coronal plane to be viewed always and for it to be simultaneously displayed with the other planes to illustrate a 3D image. With the information stored, the volumes and images can be viewed by a more experienced clinician at a later date, taking away from the operator dependency found in 2D ultrasound. (Kurjak & Chervenak, 2006). Even though these datasets can be viewed off-line making procedures shorter for patients which is the most important aspect, the clinician may have to spend more time analysing them off-line before a report can be made, making the total time for both 2D and 3D sonograms equal. (Sladkevicius, Ojha, Campbell & Nargund, 2000). Three-dimensional ultrasound has also been proven to be more precise in its measurements allowing for a superior evaluation of anatomical anomalies and an improved specificity in confirming normal pathology. (Raine-Fenning & Fleischer, 2005). Unfortunately 3D has some limitations mainly due to the modality being in its infancy. Clinicians will have to be retrained in the practice of 3D ultrasound and money will also have to be spent on new technology, proving that this is an expensive progress. Similar problems are faced in 3D as in 2D (movement, patient size, poor techniques, shadows) but over time, with more experience and new technology (Bega et al., 2003) they can be overcome to allow 3D ultrasound to replace 2D ultrasound and bridge the gap that has appeared from the use of three-dimensions in computed tomography (CT) and magnetic resonance imaging (MRI).
1.4 THREE DIMENSIONS IN GYNAECOLOGY USING OTHER MODALITIES
The use of multislice CT has increased massively since 2000, with the development of 64 detector rows in 2004. (Fischer et al., 2009). It is mainly used in cardiology but can be used for brain and abdominal scans. Its use is increasing with the possibilities of aiding surgeons performing to view a more accurate image of the internal patient anatomy before surgery. (Fischer et al., 2009). Obviously CT isn’t used very much in gynaecology due to the large amounts of ionising radiation to the reproductive area. The Royal College of Radiologists (2003) suggest that the only gynaecological or obstetric procedure to use CT would be for a suspected cephalopelvic disproportion but even this is questioned and MRI preferred. Therefore there is not much use for 3D CT especially in child-bearing age women. Three-dimensional MRI can be used for the evaluation of the hepatic and cardiac systems as well as for the assessment of the nervous system. Three-dimensional MRI has been used recently to assess foetal brain anatomy. This has had problems in the past with the movement of the foetus, and the mother and sometimes the foetus have to be sedated to provide adequate images. Fortunately now with the introduction of fast MRI sequences sedation is not needed and the brain can be evaluated three-dimensional. (Kim et al., 2010). MRI is sometimes used for the assessment of uterine fibroids, suspected pelvic masses and infertility (RCOR, 2003) but as ultrasound is the modality of choice for all these pathologies the progression of gynaecology in 3D MRI has been slow. MRI may have no ionising radiation exposure to the patient but it is a very expensive and time consuming procedure. Compared to CT and MRI, 3D ultrasound is still very much in its early stages and maybe within a few years will be become as valued as 3D CT and MRI.
2.0 SEARCH STRATEGY
2.1 JUSTIFICTATION FOR REVIEW
Justification is required for any review of literature, to access whether it is really needed. (NHS CRD, 2009). This step should start by searching through some of the relevant medical databases, namely Medline, CINAHL and PubMed. Once it has been established than no other review covers the same topic or if one does exist that it is needs to be updated, then the rest of the review process can begin. (NHS CRD, 2009). It can be said that for every 5-6 articles written about obstetric three-dimensional ultrasound only 1 concerning gynaecology has been published. (Timor-Tritsch & Monteagudo, 2007). A few reviews in some of the areas of gynaecological ultrasound already exist but they were nearly all written more than four years ago and even within the time frame that they covered, they had missed some relevant literature. This review would aim to breach that gap in an updated form.
2.2 LITERATURE REVIEW
Fortunately now with the existence of electronic databases accessible via the internet, it has made literature searching much easier. (Bowling & Ebrahim, 2005). Keywords are used to search within the databases to find the relevant literature. This in turn can provide too much information and a strict inclusion criteria has to be determined. The databases that have been used are listed in Appendix C, along with the number of articles returned in each search within the databases.
2.2.2 Peer-Reviewed Journals
It is important that all articles used have been published in peer-reviewed journals. The reason for this is so that each article has been closely examined and dissected completely to evaluate the findings in the material and verify that it is valid and original. (Bornmann & Daniel, 2008). This is like a quality control system for manuscripts that almost guarantees the reader genuine information from that article. Obviously peers are sometimes biased towards certain topics and can be shown to be human after all, when irrelevant material is published. (Bornmann & Daniel, 2008)
The keywords (see appendix C) chosen for the literature searches were:
“three-dimensional” OR 3D endometrial AND (cancer OR carcinoma) gynaecology OR gynecology polycystic OR PCOS ultrasound endometrial AND (IVF OR infertility) “tubal patency” “endometrial blood flows” ovarian AND (cancer OR mass)
These words were combined as shown using Boolean operators:
AND – includes both keywords OR – includes either one or other keyword NOT – excludes certain words from your search The use of “ “ speech marks to indicate phrases such as “tubal patency” were also used.
In an ideal world all literature should be reviewed from around the world in different languages. This would prevent a language bias into the study. The reason for the bias is that non-English manuscripts that highlight some significance in their results have a higher probability of being published in English, leaving all the articles that perhaps don’t show any significance to be left to be printed locally or not at all. (NHS CRD, 2009). This leaves the English speaking reader into thinking that all the material shows some significance, when many articles could have been rejected for the total opposite. Including all languages in this review would have been the optimal design but due to a lack of time and translation skills and facilities, it has been decided to review English only publications.
2.2.5 Inclusion Criteria
From the databases searched, many articles came up in the results. From these it had to be decided which material was to be included in the review. (Bowling & Ebrahim, 2005). To bring these numbers down, usually a time period is put as a parameter to show all publications within these dates. Due to the nature of this study, restricting the time frame could have potentially prevented relevant literature from being found as it is a fairly new subject still in its infancy. From the exclusion of non-peer reviewed publications and any poor quality reviews, a smaller list was obtained. From this list, the abstracts were read and some more irrelevant articles were filtered. Finally a more approachable list of material was allocated and it was from this literature that this review was written. (See appendix C for details of number of articles returned in a search).
2.2.6 Comparison of Literature
From the literature obtained it is important that the right information is extracted. This information can be put into a simplified table showing the key information found in each study and this will clearly show any similarities in the results between the articles. (NHS CRD, 2009). (See appendix D for the comparison tables of literature).
2.3 AREAS WITHIN GYNAECOLOGICAL ULTRASOUND TO BE REVIEWED
From the literature, it was clear that three main areas existed within the field – gynaecological cancers, fertility treatment and polycystic ovary assessment. This has been broken down even further to provide five main subject areas which this study is based on.
Critical Discussion of Literature
3.0 CRITICAL DISCUSSION OF LITERATURE
3.1 ASSISTED REPRODUCTIVE TECHNOLOGY
3.1.1 Endometrial and Subendometrial Changes
It is estimated that worldwide the average current fertility rate is 9%. (Boivin, Bunting, Collins, & Nygren, 2007). Of those couples that then decide to go on and try some form of assisted reproductive technique (ART) only 10-15% of embryos successful implant themselves in the uterus. (Salle et al., 1998). The desire to create the perfect environment within the uterus, to enable a successful embryo implantation is immense, thus encouraging many studies in the subject. Many parameters including endometrial thickness, endometrial volume and vascular indices (VI – vascularisation index, FI – flow index and VFI – vascularisation flow index) from both endometrial and subendometrial areas have been used to determine what is commonly known as uterine receptivity. “The term ‘uterine receptivity’ refers to a state when endometrium allows a blastocyst to attach, penetrate and induce changes in the stroma which result in embryonic implantation.” (Raga, Bonilla-Musoles, Casañ, Klein, & Bonilla, 1999, p.2851). From studies carried out with fertile women with regular menstrual cycles (26-32 day cycles with 3-8 days of menstrual bleeding and only mild menstrual cramping), vascular changes can be noted within the endometrium. (Jokubkiene, Sladkevicius, Rovas, & Valentin, 2006a). The follicular phase of the menstrual cycle experiences the most changes in the endometrium, when vascular indices VI and VFI, and endometrial thickness and volume increase towards the end of the phase approaching ovulation. (Raine-Fenning, Campbell, Clewes, Kendall, & Johnson, 2004; Raine-Fenning, Campbell, Kendall, Clewes, & Johnson, 2004b). The ideal method of establishing endometrial changes occurring throughout the menstrual cycle would be to take an endometrial biopsy specimen, which has proved to be successful in providing the required information from the area. (Raga, et al., 1999). To assess these changes in women who about to undergo in-vitro fertilisation (IVF), an endometrial biopsy would be too invasive. Fortunately new developments in three-dimensional sonography have enabled it to be the perfect imaging method to assess endometrial and subendometrial receptivity. (Kim et al., 2009). Three-dimensional (3D) ultrasound allows any plane through an organ to be seen and data can be acquired to provide surface rendering or volume calculation. (Alcázar, 2006). Two-dimensional (2D) sonography does not allow the whole organ or area to be looked at and only individual vessels supplying an organ can be assessed. In 3D power doppler ultrasound, the whole vessel flow can be seen to an organ or area, which is called 3D power doppler angiography (3D-PDA). (Raine-Fenning, Campbell, Clewes, Kendall, & Johnson, 2003). Virtual Organ Computer-Aided Analysis (VOCAL) can be used with 3D-PDA to analysis and measure the endometrial blood flow and volume more accurately. (Raine-Fenning, Campbell, Clewes, & Johnson, 2003). Many studies (see summary table 1 in appendix D) have tried to establish a connection between pregnancy rates and the changes seen in the endometrium using 3D sonography. The first of these was Raga et al., (1999), who demonstrated that when the blood flow measurements were taken on the day of embryo transfer – the probability of conception was connected to the endometrial volume. Women with an endometrial volume >4ml conceived 37% of the time with only 15% with volumes <2ml and no pregnancies <1ml in volume. Unfortunately a further ten studies didn’t always agree with these findings. Perhaps the biggest reason for this is in the methodology of the studies. Rather than build upon Raga et al.’s (1999) work, the next three studies all choose a different methodology allowing no comparison. (Schild et al., 2000; Schild et al., 1999; Yaman, Ebner, Sommergruber, Pölz, & Tews, 2000). The factor affecting the methodology so much in all these studies is the day when the 3D ultrasound was performed. Ng, Chan, Tang, Yeung, & Ho (2006) stated that the best time to assess endometrial and subendometrial vascularity was in the late follicular to early luteal phase. None of the studies found performed an ultrasound in the luteal phase of the stimulated cycle during IVF treatment. Two other studies were carried out on the day of embryo transfer, the same as Raga, et al. (1999), and the only agreement between the three studies were that pregnancy was never achieved when the endometrial volume was <1ml. (Kupesic, Bekavac, Bjelos, & Kurjak, 2001; Raga, et al., 1999; Zollner et al., 2003). From Jokubkiene, Sladkevicius, Rovas, & Valentin (2006b) and Raine-Fenning, Campbell, Clewes, et al.’s (2004) work on the assessment of endometrial and subendometrial changes during the normal menstrual cycle, it was established that most fertile women’s endometrial volume ranged from approximately 2-8ml. Since subfertile women cause an increase in endometrial volume (Raine-Fenning, Campbell, Kendall, Clewes, & Johnson, 2004a) it was more than likely that the endometrial volume would be >1ml anyway to achieve a pregnancy as this would have been the closest to the norm, hence not revealing anything new. The studies (Merc©, Barco, Bau, & Troyano, 2008; Wu et al., 2003; Yaman, et al., 2000) which carried out the ultrasound on the day of human chorionic gonadotrophin (hCG) administration also revealed the same results concerning endometrial volume, but all the other works came to no conclusion in regards to this. Table 2 (in appendix D) shows the parameters measured in each of the eleven studies and very few have the same parameters on the same day as the scan. Dorn et al. (2004) and Ng, et al. (2006) both measured subendometrial VI, FI and VFI but their results contradict one another. Järvelä et al. (2005) and Merc©, et al. (2008) measured endometrial VI and again their results failed to agree. None of the results from the studies agree when allowing for the day of scan to be in included in the comparison. The studies are nine years apart and in the requirement of rationalising any clinical trial, the authors should have done extensive research to justify and validate their proposed study. (Piantadosi, 2005). It seems that no conclusion can be attained from the works and makes the reader ask where the justification was in some of the later trials when this knowledge was already surfacing. Doubt in what day to scan the patient has obviously come into question, therefore why wasn’t the patient scanned on various days as was done in some the studies relating to the normal menstrual cycles? (Jokubkiene, et al., 2006a, 2006b; Raine-Fenning, Campbell, Clewes, et al., 2004; Raine-Fenning, Campbell, et al., 2004b). Järvelä, et al. (2005) tried to achieve this but only managed to do scans on two occasions in the trial and the scan produced after hCG was carried out on a variety of days in patients proving no consistency. “The day of the ultrasound examination in these studies was chosen for logistic reasons and did not take into consideration the physiological changes of endometrial blood flow throughout the menstrual cycle.” (Ng, Chan, Tang, Yeung, & Ho, 2007, p.14). Another reason for the limited scans may be due to the nature of IVF treatment being very sensitive and stressful, (Panagopoulou, Montgomery, & Tarlatzis, 2009), therefore to ask patients to come in for more scans could have risked the patient’s absence from the trial altogether. The only study to have incorporated a contrast agent (Levovist) into their study was Dorn, et al. (2004). The blood flow indices were shown to be better displayed using contrast in the study with a significant difference (P<0.001 in VI, FI and VFI measurements) between the non-contrasted results. Even though a more accurate picture may have been seen in this trial, again no correlation was made between the results and the pregnancy rate. (Dorn, et al., 2004). VOCAL rotational software has been in use for more than ten years and was used for the assessment of endometrial volume back in the year 2000. (Rudigoz, Bory, Affif, & Salle, 2000). It was established as a simple and reproducible system to measure endometrial parameters (Salle, Affif, Bory, & Rudigoz, 2000) with more reliable results than other methods. “The rotational technique proved to be superior to the conventional one, and does appear more dependent upon the morphology of an object than its absolute volume.” (Raine-Fenning, Clewes, et al., 2003, p.290). Yet only three of the studies used VOCAL, when it was available to all, bar the first couple of trials. This could have changed the results and made them more comparable. All the studies wanted to try to demonstrate some changes in the endometrium and subendometrium using 3D ultrasound and they all proved that changes do occur and that 3D ultrasound is a reliable method to do this, more so than 2D ultrasound. (Yaman, Jesacher, & Pölz, 2003). Although, their main objective was to link these results to uterine receptivity in order to create the perfect environment for embryos to implant and increase the chances of couples having successful IVF treatment. More so than the author Dr. Nick Raine-Fenning, who has written a variety of the articles on the subject, but whose incentive may be greater than others being that he is the main gynaecological consultant at a fertility clinic in Nottingham. (NURTURE: Nottingham University Research & Treatment Unit, 2010). Raine-Fenning’s articles (Raine-Fenning, Campbell, Clewes, & Johnson, 2003; Raine-Fenning, Campbell, Clewes, Kendall, & Johnson, 2003, 2004; Raine-Fenning, Campbell, Kendall, Clewes, & Johnson, 2004a, 2004b; Raine-Fenning et al., 2003) may be bias towards finding a positive conclusion as he may have more to gain. Unfortunately for him and the others, most of the works contradicted themselves in one way or another and has still left us uncertain of the relationship between uterine receptivity and endometrium and subendometrium changes. The variations in the day of the ultrasound, patient characteristics, ovarian stimulation and techniques used were the likely reasons that the results differed so much. A standard protocol for this 3D examination is required to allow future studies to be able to eventually state whether this procedure helps to increase pregnancy rates in IVF patients. Until then the value of 3D ultrasound in assessing the endometrium is limited for the predication of pregnancy and should be used cautiously. (Alcázar, 2006).
3.1.2 Fallopian Tube Patency
Establishing a cause of infertility in women can be a long drawn out process that can cause much physical and emotional pain. Up to 35% of infertile women suffer from a tubal disorder (Ahinko-Hakamaa, Huhtala, & Tinkanen, 2007) and may require some form of diagnostic testing to confirm this. Usually a tubal patency examination is the first diagnostic test carried out when starting any investigation of female infertility. (Exacoustos et al., 2009). The gold standard for tubal patency diagnosis is laparoscopy with chromopertubation (Chan, Ng, Tang, Chan, & Ho, 2005). Even though this procedure is the gold standard it has some risks as any operational procedure has which can include mortality and morbidity. (Hamed, Shahin, & Elsamman, 2009). The cost of admitting a patient and the burden on the patient has meant that NICE (2004) guidelines recommend that laparoscopy only be performed once an x-ray hysterosalpingogram (HSG) or hysterosalpingo-contrast-sonography (HYCOSY) has been carried out and a problem has been established. HSG is an inexpensive radiographic procedure that uses a contrast medium to obtain fluoroscopic images of the contrast flowing through the fallopian tubes. (Simpson, Beitia, & Mester, 2006). Tubal patency is established by the presence of contrast spill at the fimbrial end of the fallopian tube. (Kiyokawa et al., 2000). HSG is often associated with pain and discomfort due to the contrast causing tubal spasms which can briefly obstruct the tube and cause false positives. (Exacoustos, et al., 2009). It also uses ionising radiation to the reproductive area which can be damaging when potentially testing on a fertile patient. (Hamed, et al., 2009). A more patient friendly test is two-dimensional (2D) HYCOSY, which is now commonly used in infertility departments and involves transvaginal sonography with the introduction of a positive or negative contrast medium into the uterus and fallopian tubes. (Exacoustos, et al., 2009). In some studies it shows that 74% of patients prefer HYCOSY to HSG, with HYCOSY producing less pain for the patient and also enabling the assessment of extrauterine structures. (Hamed, et al., 2009). Two-dimensional (2D) HYCOSY may seem a suitable examination to replace laparoscopy as the gold standard but it does come with problems. In most of the procedures the whole tube cannot be seen in one plane as the fallopian tube is very tortuous, making the observation of free spill at the fimbrial end of the tube difficult. (Sladkevicius, 1999). This means that the procedure is very operator dependent and often is inconclusive in its results. (Kiyokawa, et al., 2000). Three-dimensional (3D) HYCOSY is the next step to overcome the problems experienced with 2D HYCOSY, HSG and laparoscopy. 3D HYCOSY allows the three planes – transverse, coronal and sagittal to be simultaneously displayed, accurately calculating any required volumes. (Merz, 1999). From this the fallopian tubes can be visualised more easily and the whole tube can be seen. The volumes and images produced by some systems can then be stored similar to HSG, for other clinicians to view at a later date and this takes away the problem of operator dependency found in 2D HYCOSY. (Exacoustos, et al., 2009). The first study to involve 3D HYCOSY was Kiyokawa, et al., (2000) and used 3D-B mode ultrasound and a sterile saline solution as the negative contrast. (See literature summary table 3 in appendix D). Only 25 infertile women were used in the study and the gold standard used was HSG. From this first study in 3D and previous studies in 2D HYCOSY it is clear that the type of contrast used is very important. The use of a sterile saline solution is limited as “the surrounding bowel and fimbrial ends have similar echogenicity and it is not easy to visualise spillage of the saline-air mix at the distal portion of the tubes.” (Exacoustos, et al., 2008, p.325). A positive contrast agent has been shown to have better results than a negative contrast and can simulate HSG findings. (Boudghène et al., 2001). Sankpal, Confino, Matzel, and Cohen, (2001) also carried out a 3D HYCOSY study with a negative contrast agent and concordance rates were 86% with the gold standard used and it was acknowledged that if a positive contrast agent was used like Echovist then the accuracy would have been greater. It might have perhaps been assumed that from these studies, all further work in 3D HYCOSY would have involved a positive contrast but this is not the case. The reason given by Sankpal, et al., (2001) was that Echovist had only been approved for use in Europe not allowing any studies in the United States (US) to be carried out involving its use. This may explain why Echovist was not used in the US but other contrasts were available, such as Optison which has approval for its use since December 31st, 1997 before the first known 3D HYCOSY study. (United States Department of Health and Human Services, 2009). This contradicts Sankpal, et al.’s (2001) explanation of why a positive contrast was not used when others were available, while clearly from previous works it would have increased accuracy. Of the European studies carried out, all used a form of positive contrast. (Exacoustas, et al., 2008; Kupesic & Plavsic, 2007; Sladkevicius et al., 2000). Concordance rates with the gold standard in successful 3D HYCOSY patients ranged from 91-99% using a positive contrast, (Chan, et al., 2005; Kupesic & Plavsic, 2007; Exacoustos, et al., 2008) whilst a negative contrast had a 64-94% concordance rate. (Kiyokawa, et al., 2000; Sankpal, et al., 2001; Ali, Kassem, Hefny, & Amin, 2005). This demonstrates that the accuracy range is smaller and more accurate with a positive contrast. Within the seven studies found, sample sizes ranged from 15 – 116 patients (see literature summary table 3 in appendix D) and none of the investigations claimed to have had a planned sample size before the start of the trial. Sample size is of great importance for any clinical trial as too few patients would mean that an anomaly with one patient could change the results by a large percentage and not represent the true results. Schulz and Grimes (2005) state that “investigators should properly calculate sample sizes and adequately describe the key details in their published report.” (p.1348). Using the sample size formula, with the standard normal distribution table (both found in Appendix E), the sample sizes used for such trials should range between 31 and 121 women depending on the confidence level used. (90% confidence level – 31 women, 95% confidence level – 43 women, 99% confidence level – 74 women and 99.9% confidence level – 121 women). This uses a 15% margin of error to compare with HSG’s accuracy to laparoscopy of 85.2% (Horowitz, Orvieto, Rabinerson, Yoeli, & Bar-Hava, 2006). Even assuming that a 90% confidence level was used, 4 (Chan, et al., 2005; Exacoustas, et al., 2008; Kiyokawa, et al., 2000; Sankpal, et al., 2001) out of the 7 studies, used less than 31 patients. This only makes the reader assume that problems were found in allocating patients for the studies and if this was the case – why? Chan, et al., (2005) used 21 patients of which only 34 fallopian tubes were able to be assessed, out of the original 42. Due to four patients being unable to carry out the procedure successfully, the concordance rate fell from 91% to 74% when including all participants. This rate is potentially the difference between a successful and an unsuccessful study. Kupesic and Plavsic, (2007) gathered 268 women for their study of which 116 were subjected to the 3D HYCOSY procedure. The concordance rate for the 3D B-mode ultrasound was 98.3% and 99.1% for the 3D power doppler ultrasound. Both of these procedures showed a much higher accuracy and if an examination on a patient was not possible, the accuracy rate would have only dropped 0.9% per person, as a larger sample size was used, instead of Chan, et al.’s (2005) 4.8% per person. It would be unfair to discredit Chan, et al.’s (2005) work with an overall concordance rate with laparoscopy at 74%, therefore the bigger picture should be looked at and noted that a 91% concordance rate was found within the successful patients. Perhaps if a larger sample size had been used in the study then the difference may have not been so great, but then again the amount of unsuccessful patients may have been greater too. Nearly all of the studies use the fallopian tube as a unit rather than the actual patient. This embellishes the results and makes for a better read but is not a true reflection of the results. (Coppus & Mol, 2006). Ali, et al., (2005) was the only study to use the patient as a unit and still managed to achieve a 94% concordance rate with the gold standard used. Fifty patients were enrolled on the study and only a sterile saline solution was used. The study does not mention whether a single operator carried out all the sonograms, therefore due to the high concordance rates, perhaps it can be speculated that a group of experienced sonographers carried out the examinations to achieve the best results. Obviously this is just an assumption but verifies the requirement for operator information in the written study to be known especially when it is recognised that 2D HYCOSY is highly operator dependant. (Kiyokawa, et al., 2000). HSG sensitivity is 85.2% and specificity is 85.2% when compared with laparoscopy. (Horowitz, et al., 2006). Since HSG has now become the gold standard in many hospital trusts, at least before establishing a problem (NICE, 2004), it can be assumed that the sensitivity and specificity rates need to be similar or higher in 3D HYCOSY to warrant the usage of this new diagnostic test. The 3D HYCOSY studies have produced some sensitivity and specificity rates that are higher than HSG and it overcomes some of the problems encountered with HSG and 2D HYCOSY. It may be achieving positive results, but 3D HYCOSY is still in its infancy and shows by the modest literature available and the small patient exposure. With more clinical trials and larger sample sizes, 3D HYCOSY may soon be the preferred choice for clinicians and patients alike.
3.2 GYNAECOLOGICAL CANCERS
3.2.1 Endometrial Cancer
Endometrial carcinoma is the seventh most frequent malignancy worldwide and is the most common gynaecology cancer. (Amant et al., 2005). Patients are normally postmenopausal and present with vaginal bleeding, even though only 10-15% of these women will be found to have a malignancy. (Alcazar & Galvan, 2009). Other disorders like endometrial hyperplasia or polyps also produce similar symptoms and therefore a correct diagnosis is important to be able to distinguish the right treatment. (Amant, et al., 2005). The ‘gold standard’ for diagnosing a patient with an endometrial carcinoma is usually dilation and curettage (D&C) or an endometrial biopsy which is performed under general anaesthetic and it carries some risks generally associated with surgery. (Yaman et al., 2002). Both these procedures take samples of tissue from the lining of the womb so that a histological diagnosis can be made. (Alcazar & Galvan, 2009; Gruboeck et al., 1996). These methods of diagnosis are both very invasive and expensive and not always enough tissue is obtained from the procedure, potentially giving a false negative. (Langer et al., 1997). An alternative test to reduce the amount of women eventually having a D&C or endometrial biopsy would be more suitable since only 10-15% of the women will actually have a malignancy. (Alcazar & Galvan, 2009). In the last two decades, more frequently two-dimensional (2D) transvaginal ultrasound has been used to measure the endometrial thickness. Endometrial thickness tends to thicken when a carcinoma is present. Studies have shown that when the endometrial thickness is less than 4mm, the likelihood of an endometrial cancer being missed is small. (Goldstein, 2009). Goldstein (2009) states that 99.9% of endometrial cancers would be diagnosed using the 4mm endometrial cut-off value. “However, endometrial thickening is a non-specific finding that may be caused by several processes, such as cancer, polyps, hyperplasia or even endometrial cystic atrophy.” (Alcázar, Castillo, Mínguez, & Galán, 2003, p.583). Differentiating between the benign and malignant pathologies would rule out the need for unnecessary procedures on some of the women. Thus other measurements would have to be taken alongside the endometrial thickness size to produce some more informative results. Testa et al. (2004) suggested that vascular assessment of the endometrial area could give important information on the existence of a cancer and its progression, whilst Gruboeck, et al. (1996) realised that the connection between endometrial volume and cancer was unknown due to the volume being unable to be measured with two-dimensional ultrasound. Therefore with the introduction of three-dimensional (3D) ultrasound, all or some of these parameters could be calculated and provide a more suitable examination for the diagnosis of an endometrial carcinoma. Gruboeck, et al. (1996) carried out the first known study using three-dimensional sonography to measure the endometrial volume. In their study a significant difference between the malignant and benign pathologies in the volume of the endometrium was found. (See appendix D, table 4, for a comparison table of the literature). In fact apart from one study (Lieng, Qvigstad, Dahl, & Istre, 2008), all the others (Alcazar & Galvan, 2009; Gruboeck, et al., 1996; Kurjak, Kupesic, Sparac, & Bekavac, 2001; Mansour et al., 2007; Merc© et al., 2007; Odeh et al., 2007; Opolskiene, et al., 2010; Yaman, Habelsberger, Tews, Pölz, & Ebner, 2008) showed that endometrial volume increased significantly. To have such an agreeable amount of literature either proves that the measuring of endometrial volume is the sure way to diagnosing a malignancy or that some bias has crept in somewhere. Looking at the literature, all the studies apart from one, (Gruboeck, et al., 1996) was carried out in a non-English speaking country. This could bring the reader to the conclusion that as most manuscripts that were originally written in a non-English language, and show some statistical differences are more likely to be published in English. (NHS CRD, 2009). Hence, what has happened to all the literature that didn’t demonstrate positive results? A language bias could be the reason. The one study that didn’t find show any differences (Lieng, et al., 2008) used a different methodology than the others. Lieng, et al. (2008) used a contrast agent when performing the sonogram to compare results between non-contrast and contrast measurements. The mean contrast measurements of the vascular indices were considerably higher than the non-contrast values but showed no difference between the malignant and benign groups. Using contrast does enable more accurate results to be obtained (Boudghène et al., 2001) but obviously the margin of error would have been the same for both the malignant and benign pathologies in the non-contrast measurements, hence not making any difference. Lieng, et al. (2008) also had the smallest patient sample which could have been an explanation for the contradictory findings. No other study used contrast so a comparison is not possible. Endometrial thickness is established as a good predictor of endometrial malignancy yet most of these studies demonstrate that endometrial volume is a better indicator to ascertain the existence of a carcinoma. Tabor, Watt, & Wald (2002) state that endometrial volume has a 4% false positive rate, whilst D&C and endometrial biopsy, the ‘gold standards’ have a false positive rate of 2-6% and 5-15% respectively. The reason that endometrial thickness may be the preferred choice of parameter instead of endometrial volume may be due to the range of cut-off values experienced in the studies for endometrial volume. These ranged from 1.35ml to 13ml (Alcazar & Galvan, 2009; Gruboeck, et al., 1996; Kurjak, et al., 2001; Mansour, et al., 2007; Merc©, et al., 2007; Opolskiene, et al., 2010; Yaman, et al., 2008). The lowest cut-off point of 1.35ml (Mansour, et al., 2007) would have shown lots of overlap in some of the other trials and the conclusion of many false positives. It is well known within the field that that any hormone replacement therapy (HRT) can change the thickness of the endometrium. It can cause vaginal bleeding and increase the risk of endometrial cancer. (Christodoulakos et al., 2006). Most of the reviews excluded patients currently taking HRT at the time of the study (Alcazar & Galvan, 2009; Gruboeck, et al., 1996; Mansour, et al., 2007; Odeh, et al., 2007; Yaman, et al., 2008). Yet none of the studies specify if HRT was used in the past. Even though a patient may not be using HRT at the time of a study, the damage may have already been done. Studies that may have had a high percentage of previous HRT users could have inflated the results in the trials and lead us to a wrongful deduction. With the common knowledge of the effect of HRT on the endometrium, perhaps it would be too hopeful that all of the studies excluded at least the patients that were currently on HRT. In fact two of the trials used a mixture of patients (Kurjak, et al., 2001; Opolskiene, et al., 2010) and they were the only two that showed a higher significance in endometrial thickness. A correlation could be made with the HRT and endometrial thickness in these studies and that would void the results. Tamoxifen, an endocrine drug used to treat breast cancer, can also effect the endometrium and not just whilst the patient is being prescribed it. Menada et al. (2004) states that 42% of women report abnormal endometrial thickness after 30 months of not taking tamoxifen. Only one of the studies excludes patients using tamoxifen (Alcazar & Galvan, 2009) but it only eliminates women who are currently using it. Since tamoxifen can have such a huge affect on the endometrium, surely all patients should be excluded from these types of studies if they have at any point used it. Alcazar & Galvan (2009) must have noted the importance of the drug in their studies by excluding current users but failed to see what the long term effects could be on their patients. Lieng, et al. (2008) noted the importance of measuring vascularisation flow, stating that “angiogenesis plays an essential role in tumour growth” (p.935), yet failed to show any significance in these parameters. These results contradict some of the other studies (Alcazar & Galvan, 2009; Merc©, et al., 2007; Odeh, et al., 2007; Opolskiene, et al., 2010) who all show significance differences in vascularisation index (VI) and vascularisation flow index (VFI). Only half of the literature found, (Alcazar & Galvan, 2009; Lieng, et al., 2008; Merc©, et al., 2007; Odeh, et al., 2007; Opolskiene, et al., 2010) measured the vascular flow indices and therefore no assumptions can be made regarding their validity. This area is still very much in its infancy and requires further studies to incorporate vascular flow indices into their methodology. Most of the studies showed that an advantage is given when using 3D ultrasound for a diagnostic tool in suspected endometrial cancer patients. Kurjak, et al. (2001) demonstrated that using 3D ultrasound provided far superior results (sensitivity of 89% for 3D v 67% for 2D and specificity of 97% for both) with only one false negative. Dr. Asim Kurjak is an established gynaecologist but recently doubt has arisen with the authenticity of his work. Kurjak is suspected of plagiarism, with his work currently being investigated, and if found to be plagiarised may be withdrawn from publication. (Chalmers, 2006). This brings uncertainty to anything that has his name and makes the reader bias towards the results in his publications, leaning away from what may be written. This takes away from what could have perhaps been an interesting and progressive development in the diagnosis of endometrial cancers. Unfortunately, with so many different cut-off rates and parameters overlapping between the studies, no definitive conclusion can be made to assess the value of endometrial volume and vascular indices in regard to the evaluation of endometrial carcinomas. Hulka, Hall, McCarthy, & Simeone (1994) suggested studying the echotexture of the endometrium along with the endometrial thickness to provide some more clear results. Kurjak, et al. (2001) used a scoring system and demonstrated that 3D ultrasound was better for evaluating the morphology of suspected malignancy. The 5 year survival rate ranges from 25-85% depending on the stage of the tumour (Amant, et al., 2005), therefore a correct diagnosis is vital to ensure the right treatment as soon as possible. Before any decision can be made on the value of 3D ultrasound in this area, more studies have to be carried out to define a clear cut-off rate for some of the parameters, using patients who have never used HRT.
3.2.2 Ovarian Cancer
“Ovarian cancer is the seventh most common cancer in women worldwide. Ovarian cancer is the fourth most frequent cause of cancer death in women and accounts for 5% of all cancer deaths.” (Laban et al., 2007, p. 201). The reason that the death rate is so high is because the illness generally produces no symptoms in the early stages that are worrying to the patient. Therefore the cancer is only found when it’s in its later stages and harder to cure having a 5-year survival rate of only 20-30%. (Kupesic & Plavsic, 2006). As the disease is asymptomatic, a screening method for women who are at high risk would be desirable but unfortunately this is difficult to do as no non-invasive test that is high enough in specificity is available. (Nossov et al., 2008). Even when a mass is found, 70% of these turn out to be benign. (Bonilla-Musoles, Raga, & Osborne, 1995). Differentiating between benign and malignant tumours has so far been very difficult to do. Conventional B-mode sonography with colour or power doppler has recently been used to aid in distinguishing between malignant and benign tumours, but some benign tumours have similar characteristics to the malignant ones, for instance cystadenofibroma or solid benign tumours. (Alcázar & Rodriguez, 2009). In fact, Guerriero et al. (2007) found that 22% of masses come under this category, where they become indistinguishable by use of ultrasound and further analysis is required. These patients who are found to have an ovarian mass are then scheduled to undergo surgery for the removal of the mass and further laboratory testing for malignancy. For an improved preoperative evaluation of the mass, certain parameters concerning the morphology of the mass can be put together to provide a more detailed and accurate diagnosis. (Geomini et al., 2006). This has proved to be subjective as no standardised parameters exist and not allowing studies to be compared. (Nossov, et al., 2008). With the introduction of three-dimensional (3D) power doppler sonography, volume box assessment of the tumour was carried out with mixed results. Half of the studies (Alcázar & Castillo, 2005; Alcázar, Galán, García-Manero, & Guerriero, 2003; Guerriero, et al., 2007; Hata et al., 1999) demonstrated that no difference was seen in the discrimination between benign and malignant ovarian masses, whilst the other half (Bonilla-Musoles, et al., 1995; Cohen, Escobar, Scharm, Glimco, & Fishman, 2001; Kurjak, Kupesic, Sparac, Prka, & Bekavac, 2003; Laban, et al., 2007) agreed that this method was good and more sensitive than two-dimensional (2D) sonography but could not be used solely for the detection of malignancy. This makes this method irreproducible and provides uncertainty in its accuracy. Lieng et al. (2008) as mentioned previously realised the importance of vascularisation within a tumour and recently this has been used to assess ovarian tumours for malignancy with 3D power doppler ultrasound. Alcázar, Merc©, & García Manero (2005) was the first to investigate this new technique. Forty-five women with sonographic adnexal masses were recruited for their study. Alcázar, et al. (2005) ended up showing that 82% of the masses in the study were malignant. Since Bonilla-Musoles, et al. (1995) states that 70% of these masses are benign, then something in the methodology was selecting women that were more likely to have malignant tumours. Alcazar’s other study (Alcázar & Rodriguez, 2009) also showed a high malignancy rate of 74%. All the other studies (Geomini, et al., 2006; Jokubkiene, Sladkevicius, & Valentin, 2007; Kudla et al., 2008) demonstrated a range of 25-34% of malignant tumours, which fits in with what Bonilla-Musoles, et al. (1995) claimed. In Alcazar’s studies the women that were chosen with adnexal masses all had central vessel distribution, which Kudla, et al. (2008) describes as the most vascular area. Alcázar, et al. (2005) actually states that tumours with central vessel distribution are more likely to be malignant, therefore their study purposely selected patients that were more likely to have malignant tumours. By selecting more women that actually had ovarian cancer then the more likely that they would get positive results as there was less to discriminate from. The importance of cystic areas within a mass has also been ignored, when (Kudla, et al., 2008) clearly stated that “cystic non-vascularised volumes of a tumour will decrease the average vascularity values, making in some cases as low as the normal physiological values.” (p.430). Alcázar, et al. (2005) obviously knew about this as their study omitted cystic areas from its evaluation, but Alcazar’s later study (Alcázar & Rodriguez, 2009) which is after Kudla, et al. (2008), includes cystic-solid masses and yet concludes later in the article to exclude cystic areas. Perhaps Alcázar & Rodriguez (2009) wanted to try and balance the results, being that the probability of a high malignancy rate was virtually guaranteed by their selection procedure. This makes all the results from the studies, apart from Alcázar, et al. (2005), who did omit cystic masses, very subjective. The results could have been more conclusive and proved to be of diagnostic value. Miyazaki et al. (1998) states that ovaries become more vascular in the luteal phase of the menstrual cycle. This was later backed up by Hope et al. (2009). All the studies (Alcázar, et al., 2005; Alcázar & Rodriguez, 2009; Geomini, et al., 2006; Jokubkiene, et al., 2007; Kudla, et al., 2008) include women that are pre and post menopausal but only one of them Kudla, et al., (2008), who has also worked with Hope et al., (2009) took this into consideration and scanned the premenopausal women in the follicular phase. Kudla, et al. (2008) showed that the vascular indices showed significant differences in the follicular and luteal phases of the menstrual cycle but more significance was found in the luteal phases which agrees with Hope, et al. (2009) and Miyazaki, et al. (1998) and suggests that their analysis was correct. The other four studies (Alcázar, et al., 2005; Alcázar & Rodriguez, 2009; Geomini, et al., 2006; Jokubkiene, et al., 2007) obviously knew about this but choose not to include it in their methodology. By excluding this from their study, the vascular indices could be embellished if the scan was taken in the luteal phase. The contraceptive pill is known to decrease the risk of ovarian cancer (Hope, et al., 2009) but yet only one of the works (Jokubkiene, et al., 2007) took a full history of the women including their use of the contraceptive pill and hormonal replacement therapy drugs. Hope, et al. (2009) states that “hormonally suppressed ovaries were significantly less vascular,” (p.1049). Therefore women that may have been taking any of these medications could have skewed the results and weakened any conclusions that arose. Jokubkiene, et al. (2007) also used a 5cm3 spherical sample of the tumour, as well as the whole tumour, meaning that the area that was mostly vascularised was used. This originally was used so that none of the actually ovary was used to be tested but Jokubkiene, et al. (2007) agrees that the selection of these 5cm3 spherical samples is subjective to the clinician and not reproducible. Some tumours are smaller than 5cm3, so the question is how did Jokubkiene, et al. (2007) include all the tumours in the study. Kudla, et al. (2008) used the same strategy but used a 1cm3 spherical sample to be able to include all tumours. Despite all the failings in these studies the subject area is still in its infancy, with only five studies being carried out and all of them showed some significant differences in their vascular results especially in the flow index (FI). Alcázar & Rodriguez (2009) showed that this new technique decreased the false positive rate by 33% without lowering sensitivity, this would mean that the 22% of indistinguishable tumours (Guerriero, et al., 2007) would decrease to 14.6%. Jokubkiene, et al. (2007) agrees that some 10% of tumours will be hard to diagnose leaving 10 in 100 patients unsure of their outcome. The specificity in these results needs to be high so that patients are not assessed as false negatives and potential ovarian cancer patients could be told that their mass is benign. Geomini, et al. (2006) showed the highest specificity with 85% but only 57% sensitivity, this concludes that this technique alone for the discrimination between benign and malignant tumours is not viable and further study has to be carried out with larger samples to determine its true benefits. Kupesic & Kurjak (2000) used contrast in their 3D power doppler volume box assessment of adnexal masses and found a sensitivity rate of 100% and specificity of 93.9% which produced better results than any of the five (Alcázar, et al., 2005; Alcázar & Rodriguez, 2009; Geomini, et al., 2006; Jokubkiene, et al., 2007; Kudla, et al., 2008) 3D power doppler angiogram studies using virtual computer-aided analysis (VOCAL) which is supposedly more accurate. (Raine-Fenning, Campbell, Clewes, & Johnson, 2003). Kupesic & Kurjak (2000) suggest that their results would have increased if morphological parameters had been taken into account, which Geomini, et al. (2006) also agreed. Thus further study should include the considerations mentioned above as well as take morphology and perhaps contrast into account.
3.3 EVALUATION OF THE OVARIES
3.3.1 Polycystic Ovary Syndrome
Polycystic ovary syndrome (PCOS) is the most common endocrine disorder that affects pre-menopausal women. (Speca, Napolitano & Tagliaferri, 2007; Balen, 1999). Various studies have been carried out to try and assess the prevalence of PCOS in the population and s range from 17-22%, with 22% being the most common. (Hart, Hickey & Franks, 2004). There is no known cause of the syndrome and its pathogenesis varies from literature to literature, hence the lack of international definition. (Speca et al., 2007). Until 2004, the most recognised definition of PCOS came from the National Institutes of Health (NIH) 1990 criteria for PCOS, and this described that patients who had PCOS had to have: hyperandrogenism and/or hyperandrogenemia, oligoovulation, and the exclusion of any other related disorders. (Azziz, 2006). As these criteria did not include the diagnosis of polycystic ovaries using ultrasonography, it was thought to be controversial. The polycystic ovary should have 12 or more follicles present measuring 2-9mm in diameter or for at least one of the ovaries to have a volume of 10cm3. (Balen, Laven, Tan & Dewailly, 2003). Therefore in 2003, a meeting between the European Society of Human Reproduction and Embryology and the American Society for Reproductive Medicine produced new criteria for the definition of PCOS, which was called the Rotterdam 2003 Criteria and to this day it continues to be the ‘gold standard.’ It had been clear for some time that PCOS had a wide range of symptoms and the new criteria is thought to encompass more variations of the syndrome. Patients diagnosed with the old NIH criteria would also meet the new Rotterdam 2003 criteria. (Azziz, 2006). Symptoms in PCOS can include oligoovulation, anovulation, signs of hyperandrogenism which can be hirsutism, acne, seborrhoea, alopecia and obesity, and polycystic ovaries. (Speca, 2007). Women also find themselves having fertility problems due to an irregular or no menstrual cycle. (Ng, Chan, Yeung & Ho, 2005). The new definition would first exclude any other disorders that may present in a similar way to PCOS and then the patient would expect to have two out of three of the following features: “oligo- or anovulation, clinical and/or biochemical signs of hyperandrogenism, or polycystic ovaries.” (Rotterdam ESHRE/ASRM Sponsored PCOS Consensus Workshop Group, 2004, p.782). It must be made clear at this point the difference between PCOS and polycystic ovaries (PCO). Polycystic ovary patients present just with polycystic ovary morphology and none of the other symptoms associated with PCOS. (Rotterdam ESHRE/ASRM Sponsored PCOS Consensus Workshop Group, 2004). Transvaginal (TVS) ultrasonography has long been the modality of choice to image the ovaries and uterus and allows a non invasive assessment of the area. The interest in studying PCO and PCOS solely has increased over the last few years mainly due to the desire to boost success rates in IVF procedures (Timor-Tritsch & Monteagudo, 2007) and various studies have been carried out describing the features of PCO on three-dimensional ultrasound. Kyei-Mensah, Tan, Zaide & Jacobs (1998) were among the first to use 3D ultrasonography in assessing patients with PCOS. Measurements were taken in the early follicular phase (days 2-5 of the menstrual cycle) as many patients were due for IVF treatment in the late follicular phase. (See appendix D for literature comparison tables). Most of the literature (Lam, Johnson & Raine-Fenning, 2007; Lam, Raine-Fenning, Cheung & Haines, 2009; Mala, Ghosh & Tripathi Ng et al., 2005; Pan et al., 2002; Pascual et al., 2008) agreed with Kyei et al., (1998) and scanned patients in the early follicular phase (between days 2-5). Balen et al. (2003) states that if follicles are >10mm then the scan should be repeated when ovarian quiescence is occurring, which is likely to be during the early follicular phase. Larger follicles are more common later in the follicular phase and can give a false diagnosis of PCO. Unfortunately Dolz et al. (1999) and Jarvela et al. (2002) didn’t recognise the need for scans to be performed early on in the follicular phase, scanning patients in the late follicular phase instead and their results are difficult to compare with others and may show a higher degree of change within the ovaries due to the larger follicles in the late follicular phase. Unfortunately due to this questionable methodology (Dolz et al., 1999; Jarvela et al., 2002) their results are weakened. Jarvela et al. (2002) used patients with PCO rather than PCOS which all the patients in other studies had. This excluded all patients with clinical symptoms of PCOS. Hyperandrogenism is an unpleasant symptom of PCOS but not PCO, and the degree of hyperandrogenism has shown to be important, seen in the later studies affecting the outcome of the results. Normal weight PCOS patients seemed to increase VI and VFI, (Lam, Johnson & Raine-Fenning, 2007; Lam, Raine-Fenning, Cheung & Haines, 2009; Ng et al., 2005; Pan et al., 2002) but yet hyperandrogenism sufferers are more likely to have an increased body mass index (BMI) suggesting that the smaller BMI patients have a less severe case of hyperandrogenesism. (Dolz et al., 1999). This may explain why the studies involving Chinese patients all demonstrated significant differences in vascular indices, (Pan et al., 2002; Ng et al., 2005; Lam et al., 2009) due to Chinese women having a lower prevalence of hyperandrogenism. (Lam et al., 2009). The amount of hirsute women showed no correlation between VI and VFI again demonstrating the effect of hyperandrogenism on the vascularisation of the ovaries. (Lam et al., 2007). Jarvela et al. (2002) by ruling out PCOS women, really put the study in a different category and the results have become invalid when trying to make correlations with other studies. The vascularisation of the ovaries has been shown to be important and the results showed this. Kyei-Mensah et al. (1998) did not measure vascular indices, therefore excluding Kyei-Mensah et al. (1998), and Dolz et al. (1999) and Jarvela et al. (2002) for scanning patients in the late follicular phase, all other studies apart from Pascual et al. (2008) showed some form of increase in the vascular indices. Aleem & Predonic (1996) state that within PCOS there are also a variety of subcategories (phenotypic manifestations) where ovarian vascularity may differ. These subcategories have to be considered when interpreting results and future consideration of these subcategories should aid in the diagnosis of PCOS. Lam et al. (2007), Lam et al. (2009) and Ng et al. (2005) saw the importance of the different phenotypic manifestations within PCOS patients and used them as a tool to aid the diagnosis of PCOS. Pascual et al. (2008) and Mala et al. (2009) obviously saw the significance of these different clinical and laboratory manifestations, by evaluating BMI, ovulation and hirsutism, but they were not taken into consideration when analysing their results, making these studies potentially incomplete. By then, so much was already known about the effect of hyperandrogenism in PCOS that to totally ignore a relationship between the two, really takes away from the credibility of the study and the question could be asked why the trial was justified to begin with. Five of the studies (Jarvela et al., 2002; Kyei-Mensah et al., 1998; Lam et al., 2007; Lam et al., 2009; Pan et al., 2002) recruited women that were receiving fertility treatment. It is unknown what the reasons for their infertility but some factors can give us a false impression. For example – problematic ovarian blood flow is a cause for infertility but it could have also inflated the vascularisation in the ovary. (Lam and Raine-Fenning, 2006). Fertility drugs given to patients to stimulate their ovaries ready for ovulation can result in ovarian hyperstimulation syndrome (OHSS), which sometimes produces enlarged ovaries. (Villasante et al., 2008). This could have enhanced the results in these studies and produced a significant result when actually there may not have been one if the patients had been selected at random. Recruiting women from fertility clinics brings into question the selection technique and design study. The authors may have already known and examined the patients and selected the ones that they thought might give the best results. (Lam & Raine-Fenning, 2006). All the studies demonstrated an increase in ovarian volume in the PCOS patient which was to be expected as a polycystic ovary has more follicles, making its stroma volume greater and increasing the overall volume. (Balen et al., 2003). Although this could have just been due to the infertile patients used perhaps being on ovary stimulating medication. In conclusion these nine studies all provided interesting results but were difficult to compare due to dissimilar methodologies and different patient characteristics. Further larger scale studies need to be carried out to give definite parameters for vascular indices, which can include the assessment of BMI, ovulation and hirsutism on the PCOS patient in 3D ultrasound. It would also help if previous works were taken into consideration when trying to create a methodology for a new study incorporating the conclusions from older works. This will “assist in defining the severity, progression or regression of the disease.” (Mala et al., 2009, p.38).
Conclusions and Further Recommendations
4.0 CONCLUSIONS AND FURTHER RECCOMMENDATIONS
4.1.1 Endometrial and Subendometrial Changes in IVF
The desire to enhance pregnancy rates in in-vitro fertilisation (IVF) for both financial and scientific reasons has led many authors to concentrate on the endometrial and subendometrial changes that might occur when carrying out IVF. Unfortunately with this desire to succeed, an immense rush to carry out trials has occurred without any consistency. The objective of any trial is to rule out a theory or to affirm a statement. Normally works are compared with others to emphasis and strengthen conclusions but in this case each author had their own agendas and after eleven trials no real conclusions can be made. Vascular changes do occur in the endometrium and subendometrium in IVF but to what extent is uncertain and a call for a standard protocol to include information such as patient characteristics, day of scan, technique used, equipment, evaluation and subendometrial region is required to be able to ascertain whether this actually does help in IVF. (Ng et al., 2007).
4.1.2 Fallopian Tube Patency
Two-dimensional (2D) hysterosalpingo-contrast-sonography (HYCOSY) is already common practice in many gynaecological departments. (NICE, 2004). Three-dimensional (3D) HYCOSY has shown to be of equal diagnostic value or better (see appendix D, table 3) and therefore it should replace 2D HYCOSY as it demonstrates benefits not achieved with 2D HYCOSY. These include less exposure time with the patient, reduced contrast required – so not as painful for the patient, (Kupesic & Plavsic, 2007), and it eliminates the problem of 2D HYCOSY operator dependency when the volumes can be stored online for further assessment at a later date. (Chan et al., 2005). The only hold back for 3D HYCOSY is that it requires clinicians to be retrained and new equipment capable of scanning in 3D to be bought. All of this requires financial expenditure and may put many trusts off preferring to use 2D HYCOSY or x-ray hysterosalpingogram (HSG) as an alternative cheaper method. (Chan et al., 2005).
4.1.3 Endometrial Cancer
Endometrial volume in most cases achieved better results than the currently used endometrial thickness evaluation for determining a malignant endometrium. Unfortunately with no standard protocol and cut-off rates in the studies being all different with overlaps, no definite conclusion can be made. The studies show promise and perhaps with some standardisation and the implementation of a scoring system for the morphology and the endometrial volume and thickness as suggested by Kurjak, et al. (2001). This could be a new procedure to cut down the necessity of surgery and eliminate women from an unnecessary endometrial biopsy or dilation and curettage (D&C) to determine malignancy.
4.1.4 Ovarian Cancer
Preoperative assessment of ovarian masses has always been subjective depending on many factors. Ovarian cancer is very aggressive once found in its later stages. When masses are found to be benign, clinicians still prefer them to be removed requiring surgery. Therefore a more fitting use for 3D ultrasound would be for a screening process that could determine the early stages of cancer in women. This requires the procedure to be sensitive enough to capture the right information early on. Finding a suitable examination to do this has been difficult with conflicting results (Nossov, et al., 2008) but Kupesic & Kurjak (2000) and Geomini, et al. (2006) both merit the use of 3D ultrasound in this area and suggest that morphological parameters also be used to provide more accurate results. More work has to be done in this area as it is still early days, to be able to come to definite conclusions.
4.1.5 Polycystic Ovary Syndrome
Some of the later studies showed promise but then when analysing their data failed to review previous author’s failings to improve their own work. This was rather disappointing and shows the need for justification in carrying out trials, to verify their benefit. Again, some form of standard protocol is required to allow conclusions to be made on 3D ultrasounds usefulness in this area. It is clear that the day of the scan and the degree of hyperandrogenism in a patient is so important and this needs to be included in every trial. For something that is prevalent in 22% of women, (Hart et al., 2004), the sample sizes were very small and mostly women who were carrying out fertility treatment were recruited. This could lead to a biased conclusion being that the women may have had other problems associated with infertility. A very enjoyable but frustrating read none the less.
One thing that is clear from all the studies done using 3D ultrasound is the lack of standardisation and protocol. Many would have benefitted from some form of standard protocol and allowed for study to study comparisons. The chosen techniques were sometimes random and unexplainable when previous authors had proved otherwise. The sample sizes of the trials also ranged from 15 to 451 women demonstrating no statistical evidence in how they came to the sample size that they used. A sample is supposed to represent a mathematical proportion of a group in a society and needs to be sufficient to allow conclusions to be made from the study. In all, some very interesting work that requires further study to reveal the true benefits of 3D sonography.
4.2 RECOMMENDATIONS AND FURTHER STUDY
Standardisation for the methods used in the five areas of study needs to be devised so that further conclusions can be made from new works. This would mean reviewing all the current literature establishing what parts are important in the methodology, technique, patient characteristics and scan schedules. Cut-off values for the parameters used need to be set using an appropriate sample size that reflects the proportion of patients affected by the pathology. Most of these studies proved that the parameters used on their own may not solely make a diagnosis, therefore other parameters such as morphology and patient characteristics need to be taken into consideration. A standardisation in these areas would allow studies to be compared more successfully and could accelerate the use of 3D ultrasound in a department if a diagnostic value is proved. Some of the studies have already mentioned that a scoring system could be used to enhance 3D ultrasound results. Once a standard protocol is devised for the scoring system, then this could be used especially in the gynaecological cancer studies where the outcome is so important in the early detection of cancer. Kurjak, et al. (2001) included information like vessel architecture, septa evaluation, echogenicity and branching pattern in their scoring system but admited that even though only a small proportion of studies used this method none of them provided the same scoring system. It is abundantly clear that without a scoring system in some of the areas that further progress cannot be made.
Reflection and Self-Evaluation
5.0 REFLECTION AND SELF EVALUATION
I never realised that there was so much literature already available on a technology that is fairly new. Time became an important factor when doing this dissertation. Even though, the actually writing part didn’t take as long, the copious amounts of literature to be read and evaluate took so much time. As there was so much, I had to categorise the literature into the areas that I was going to be looking at and read each section separately, whilst writing notes. Once I had read each area, I would write up about this topic for my dissertation. This meant that I didn’t get confused with the topics and cross-referenced irrelevant material. I should have done this straight from the start instead of trying to read everything together and getting bamboozled with so much information. Even though I may have criticised my time management, I am still writing this reflection over one week ahead of the due date, with everything else done, so I don’t think it has gone that badly! It was frustrating reading some of the literature as certain things may have been expected from the authors and they didn’t deliver. This made me realise that clinicians are fallible and it is fine to criticise their work, which to begin with I felt unqualified to do so. This made it difficult in trying to be critical of the works and instead being descriptive. It took time to get there and understand the concept but eventually I got there. This was very much a piece of work that I enjoyed and benefitted, emphasising more so my desire to further my knowledge in ultrasound. It has been a long road and I am glad that I can now look back and write this reflection with the knowledge that I still have some time left!
Assessment Max mark My Mark Comments Abstract/Introduction/Background 10 5 The introduction was especially difficult with an uncertainty of how much depth to go into. Search Strategy 10 6 This was more time consuming than anything and a bit monotonous. Critical Discussion 50 26 Hard to get the right balance of descriptive to critical discussion. Conclusions & Recommendations 10 5 So many conclusions could have been made that it was hard to keep it to the point. Presentation and Referencing 10 7 The help of Endnote made the laborious referencing easier. Management of dissertation 10 6 Starting off slow but finished well in time. Total 100 55 53