- Top of page
Objective The objective of this study was to investigate whether follicle stimulating hormone (FSH), anti-Mullerian hormone (AMH) and inhibin B could be useful in predicting the ovarian response to gonadotrophin stimulation in assisted reproduction patients who are considered to be poor responders.
Design Prospective study.
Sample Blood samples were collected on day five or six in the early follicular phase of an untreated menstrual cycle. Samples were collected from 69 patients.
Methods Serum samples were assayed for FSH, AMH and inhibin B using commercial immunoassay kits.
Main outcome measures Response to gonadotrophin stimulation and number of eggs collected.
Results Among the 69 patients, 52 patients completed an IVF cycle and 17 patients had to cancel the cycle because of poor ovarian response to gonadotrophin stimulation. Mean FSH levels were significantly higher (P < 0.05) in the cancelled group (10.69 ± 2.27 mIU/mL) compared with the cycle-completed group (7.89 ± 0.78 mIU/mL). Mean AMH levels were significantly lower (P < 0.01) in the cancelled group (0.175 ± 0.04 ng/mL) compared with the cycle-completed group (1.13 ± 0.2 ng/mL). Mean inhibin B levels were significantly lower (P < 0.001) in the cancelled group (70 ± 12.79 pg/mL) compared with the completed group (126.9 ± 8.8 pg/mL). Predictive statistics show that AMH is the best single marker and that the combination of FSH, AMH and inhibin B is modestly better than the single marker. Linear regression analysis in the cycle completed patients shows that although FSH (r= 0.25, P < 0.05) and inhibin B (r= 0.35, P < 0.05) have a significant linear association with the number of eggs collected, AMH has the greatest association (r= 0.69, P < 0.001) with the number of eggs collected among the parameters measured.
Conclusion In this particular group of IVF patients, AMH is the best single marker of ovarian response to gonadotrophin stimulation. The combined markers modestly improved the prediction.
- Top of page
It is known that reproductive ageing is related to reduction of the primordial follicle pool. As women's age increases, their ovarian reserve diminishes, and the rates of both spontaneous and assisted reproductive pregnancies decline. Identification of patients who are at risk of poor response to ovarian stimulation can help patients decide whether to undergo demanding fertility treatment and guide the clinician in choosing the gonadotrophin stimulation protocol. Assessment of ovarian response potential before the patient enters an in vitro fertilisation (IVF) program is therefore important. For several years, age and day three follicle stimulating hormone (FSH) levels have been used as an indicator of ovarian response in assisted reproductive technology. Recently, various studies have reported the usefulness of antral follicle count and ovarian volume in predicting ovarian response to hormone stimulation.1–4 These studies conclude that antral follicle count provides better prognostic information on the occurrence of poor ovarian response during hormone stimulation for IVF than does the patient's chronological age or basal FSH. However, ultrasonography is subjective and the interpretation of the observation may not be consistent. Therefore, there is a need for a biochemical endocrine marker that could be used in all centres without bias.
Several studies have evaluated the use of basal inhibin B in predicting ovarian response in assisted reproductive technology.5–10 The evidence so far is discordant. Recently, anti-Mullerian hormone (AMH) has been evaluated by some groups as a marker of ovarian reserve.4,11 Inhibin B and AMH are members of the TGF beta family. Ovarian follicular granulosa cells produce AMH and inhibin B. Both these hormones are produced by small antral follicles and could be useful endocrine markers of the antral follicle pool in the early follicular phase. Inhibin B levels in serum rise in the early follicular phase and peak around days five to six of the cycle12 and then gradually fall in the late follicular phase and are almost below detection in the luteal phase with a small peak in levels post-ovulation. Although cyclical pattern for AMH has not been yet reported, one study has found similar levels of AMH at different stages of the menstrual cycle.13
This prospective study was carried out to investigate the use of inhibin B, AMH and FSH on day five or six of the cycle in a group of patients who were expected to have poor response because of their age, raised day three FSH levels or previous history of poor response in an in vitro fertilisation (IVF)/intracytoplasmic sperm injection (ICSI) treatment. Our aim was to determine if these markers would significantly contribute to the prediction of a cancelled IVF cycle in this particular group of patients and to study the relationship between the number of eggs collected and these hormone levels in the early follicular phase.
- Top of page
Patients who attended the infertility clinic at Essex Nuffield Hospital between June 2002 and September 2003 were assessed and if they fulfilled one or more of the selection criteria, their day five or six blood sample was collected. Samples were collected as part of infertility investigations and these patients were managed based on accepted principles of infertility practice.
The three main selection criteria were, (i) >38 years of age, (ii) day three serum FSH >10 IU/L and (iii) a history of poor response in a previous cycle of IVF/ICSI. We included 69 patients in the study. Seventeen patients had to either convert their IVF/ICSI cycle to an intrauterine insemination or totally cancel the cycle because of poor response. A patient was considered a ‘poor responder’ if she developed less than four follicles >15 mm diameter.
These patients were treated using two protocols: (i) a long cycle with GnRH agonist down-regulation or (ii) a short cycle with a GnRH antagonist. Two types of gonadotrophins, menopur (Ferring Pharmaceuticals, Berks, UK) and gonal F (Serono Pharmaceuticals, Middlesex, UK) were used for stimulation. In the long protocol, a daily injection of buserilin (Aventis Pharma, Kent, UK) 0.5 mL was given from day one of the cycle for two weeks and once complete pituitary down-regulation was achieved based on the ultrasonographic evidence of an endometrial lining of <5 mm, daily injections of gonadotrophin commenced and the dose of buserilin reduced to 0.2 mL daily. When three or more follicles reach >18 mm, both the gonadotrophin and buserelin injections were stopped and 10,000 IU of hCG (Pregynl, Organon, Cambridge, UK) was given.
In the short protocol, daily gonadotrophin injection was commenced on day two of the cycle and once the lead follicle was >14 mm, daily injections of either cetrotide 0.25 mg (Serono Pharmaceuticals) or orgalutron 0.5 mg (GnRH antagonists) (Organon) was administered. When three or more follicles reach >18 mm, both injections were stopped and an injection of 10,000 IU hCG was given.
Egg collection was planned 34–36 hours after hCG injection, followed by embryo transfer two days later. Following embryo transfer, patients were given luteal support in the form of either cyclogest vaginal pessaries at 400 mg (Shire Pharmaceuticals Group, Hampshire, UK) twice daily or crinone gel 8% (Serono Pharmaceuticals) once every night or gestone injection (Nordic Pharma, Berks, UK) at 100 mg im daily until the day of the urinary pregnancy test 15 days after the embryo transfer.
All samples were assayed in duplicate using a commercial assay kit according to the manufacturer's sensitised assay protocol (Immunotech, Marseille, France). The sensitivity of the assay was 0.098 ng/mL. The intra- and inter-assay variations were <15% using an in-house quality control pool.
All samples were assayed in duplicate using a commercial ELISA kit (Oxford Bio-Innovations-DSL, Oxford, UK) according to the manufacturer's protocol. The sensitivity of the assay is 10 pg/mL. The intra- and inter-assay variations were <10%.
Follicle stimulating hormone
All samples were assayed in duplicate using a commercial Coat a count solid phase immuno-radiometric assay (IRMA) (DPC, Gwynedd, UK). The sensitivity of the assay is 0.06 mIU/mL and the intra- and inter-assay variations are <7%.
Initial tests were carried out to see whether there was a significant difference in the ovarian response between the treatment protocols or gonadotrophin preparations used for stimulation.
Student's t test was carried out to test the difference between the means in the cancelled and completed IVF cycles.
Diagnostic tests were performed by calculating sensitivity and specificity using the area under the ROC curve for FSH, inhibin B and AMH for the prediction of cancellation of an IVF cycle.
Regression analysis was carried out to study the relationship between the measured parameters and the number of eggs collected. In order to normalise the different number of gonadotrophin units used in different patients, the number of eggs collected was multiplied by a constant (×1000) and then divided by the total units of gonadotrophins used for the IVF stimulation before statistical analysis. STATA statistical software (Stata, Texas, USA) was used for statistical analysis.
- Top of page
The different treatment regimens (down-regulation protocols or gonadotrophin preparations) did not have a significant effect on the number of follicles developed or the number of eggs collected. Therefore, hormone measurements in patients from both treatment protocols were pooled to evaluate the objectives.
Serum concentrations of FSH, inhibin B and AMH were measured on day five of the menstrual cycle. Levels of these hormones were compared in women who had a complete IVF treatment cycle and women who cancelled their treatment cycle due to poor response to gonadotrophin stimulation. There were 52 patients who had a complete IVF cycle and 17 patients who had a cancelled cycle. There was no significant difference in the age between the cancelled (31–48 years) and completed IVF cycle patients (27–45 years). Mean FSH levels were significantly higher (Fig. 1a; P < 0.0.05) in the cancelled cycle patients (10.69 ± 2.27 mIU/mL; 2.6–41.3 mIU/mL) compared with the completed cycle patients (7.89 ± 0.78 mIU/mL; 2.3–16.7 mIU/mL). Mean inhibin B levels were significantly lower (Fig. 1b; P < 0.001) in the cancelled group (70 ± 12.79 pg/mL; 10–172 pg/mL) compared with the completed treatment group (126.9 ± 8.8 pg/mL; 10–262 pg/mL). Serum AMH levels were significantly lower (Fig. 1c; P < 0.01) in the cancelled group (0.175 ± 0.04 ng/mL; 0.098–0.63 ng/mL) compared with the IVF completed group (1.13 ± 0.2 ng/mL; 0.1–5.29 ng/mL). In patients who cancelled their IVF cycle, 4 of 17 patients had measurable amounts of AMH. However, 6/52 patients who had a complete IVF cycle also had undetectable levels of AMH in the serum. Although the mean levels of AMH and inhibin B were significantly different between the groups, there was considerable overlap.
Figure 1. Scatterplot of individual concentrations of (a) FSH, (b) inhibin B and (c) AMH in patients who had a completed IVF cycle (n= 52) and cancelled IVF cycle (n= 17). Unpaired Student's t test was carried out to compare the means. *P < 0.05, **P < 0.01, ***P < 0.001.
Download figure to PowerPoint
Sensitivity is the fraction of positive cases that are correctly classified by the diagnostic test; in this study it is the cancelled cycles. Specificity is the fraction of negative cases that are correctly classified; in this study it is the percentage of completed IVF cycles. Various cutoff points were analysed and the cutoff points that gave the best sensitivity, specificity and % correctly classified were used for analysis.
Table 1 shows the sensitivity (true positives/cancelled cycles) and the specificity (true negatives/completed IVF cycle) for FSH (Table 1a), inhibin B (Table 1b) and AMH (Table 1c) using cutoff points of 8 mIU/mL for FSH, 80 pg/mL for inhibin B and 0.1 ng/mL for AMH. FSH had a sensitivity of 77% and a specificity of 44% with ∼68% of the population being correctly classified. Inhibin B had a sensitivity of 81.25% and a specificity of 44.44% and 71.21% of the population were correctly classified. AMH had a sensitivity of 87.5% and specificity of 72.2% and 83.33% of the population were correctly classified. An arbitrary score was given in each case to perform a diagnostic test for the combined markers. The cumulative score was used to predict the diagnostic capability of all three hormones (Table 2). The cumulative score of 5 or 6 had a sensitivity of 91.67% or 62.5% and a specificity of 44.44% and 83.33%, respectively, and 68% of the population were correctly classified. The positive likelihood ratio was 3.75 for the latter. However, the results show that there is no single marker or combined markers that have 100% sensitivity and 100% specificity (Tables 1 and 2) or that could correctly classify 100% of the population.
Table 1. Predictive statistics providing the sensitivity, specificity, percentage of patients correctly classified and the likelihood ratio of a positive test for (a) FSH with a cutoff of 8 mIU/mL, (b) inhibin B with a cutoff of 80 pg/mL and (c) AMH with a cutoff of 0.1 ng/mL. The area under the ROC curve for each parameter is given with 95% confidence intervals within brackets. An arbitrary score is given for each parameter.
| ||Score||Sensitivity(%)||Specificity(%)||Correctly classified(%)||Positive likelihood ratio|
|Area under the curve (ROC area) = 0.566 + 0.06 (0.45–0.68)|
|(b) Inhibin B|
|Area under the curve (ROC area) = 0.628 + 0.06 (0.497–0.684)|
|Area under the curve (ROC area) = 0.798 + 0.06 (0.68–0.91)|
Table 2. Predictive statistics providing the sensitivity, specificity, percentage of patients correctly classified and the likelihood ratio of a positive test for the combined markers using the cumulative score obtained from the scores inTable 1. The area under the ROC curve for the cumulative is given with 95% confidence intervals within brackets. Score 3 = FSH + inhibin B + AMH (all below cutoff); Score 4 = FSH (>10 mIU/mL) + inhibin B (<80 pg/mL) + AMH (<0.1 ng/mL); Score 5 = FSH (>10 mIU/mL) + inhibin B (>80 pg/mL) + AMH (<0.1 ng/mL); Score 6 = FSH (>10 mIU/mL) + inhibin B (>80 pg/mL) + AMH (>0.1 ng/mL).
|Cumulativescore||Sensitivity(%)||Specificity(%)||Correctly classified(%)||Positive likelihood ratio|
|Area under the curve (ROC area) = 0.78 + 0.06 (0.66–0.9)|
In the patients who had a complete IVF cycle (n= 52), linear regression analysis was carried out on the relationship between the basal hormone levels and the number of eggs collected, normalised to the units of gonadotrophins used for stimulation (as described in the Methods section). FSH (r= 0.25, P < 0.05) and inhibin B (r= 0.35, P < 0.05) had a significant linear relationship with the number of eggs collected (Fig. 2a and b). Serum AMH had the most significant linear relationship (r= 0.69, P < 0.001) among the three markers explored with the number of eggs collected. Combined use of FSH, inhibin B and AMH marginally improved the relationship (r= 0.7, P < 0.001) between the three markers and the number of eggs collected.
Figure 2. Scatterplot of individual concentrations of (a) FSH, (b) inhibin B and (c) AMH in patients who had a completed IVF cycle (n= 52) against the number of eggs collected normalised to the number of gonadotrophin units used for stimulation (No. of eggs collected × 1000/total units of gonadotrophins). Linear regression lines are shown with the regression coefficient r and P value.
Download figure to PowerPoint
- Top of page
Predicting the ovarian reserve in IVF patients is useful in counselling patients as well as optimising the gonadotrophin stimulation protocol to obtain a good response. It has been reported that day three FSH, inhibin B and oestradiol (E2) are particularly useful in guiding the choice of optimal protocol for assisted reproductive technology.9 Pituitary FSH secretion increases with declining ovarian reserve and day three FSH levels have been routinely used as a marker of ovarian reserve in assisted reproductive technology.14,15 Age and FSH are indirect markers of ovarian reserve and this might explain the relative insensitivity in predicting ovarian response in IVF. Inhibin B and AMH are produced by the small follicles in the ovary and could be sensitive direct markers of ovarian reserve. FSH, E2 and inhibin B are regulated by the hypothalamic–pituitary–ovarian feedback loop.
This study was designed as a pragmatic study reflecting normal IVF unit treatment regimes and patient management. In this study, we have chosen a group of patients who were considered as poor responders to IVF treatment and would particularly benefit from knowing the chances of success of the IVF treatment. It is evident from the results that a single hormone cannot be used as a sensitive marker of ovarian reserve. FSH levels were significantly higher in the cancelled cycles, while inhibin B levels were 50% lower in the cancelled cycles compared with patients who had a completed IVF cycle. AMH levels were almost 10-fold lower in the cancelled cycles compared with patients who had a completed IVF cycle. In ∼75% of cancelled cycles, AMH levels were below detection limit. However, ∼12% of the completed IVF cycle patients also had undetectable levels of AMH. Therefore, the overlap in these data excludes the possibility of AMH being used as a single marker predicting a cancelled IVF cycle. Predictive statistics show clearly that none of the three parameters measured could be 100% predictive of a cancelled cycle.
Linear regression analysis shows that AMH has the most significant relationship with the number of eggs recovered. Combined analysis of AMH, FSH and inhibin B only marginally improves the relationship of the hormones to the number of eggs collected.
Several groups have used inhibin B as biochemical markers to predict ovarian response to IVF treatment. These studies were conducted using samples taken on various days of the untreated and stimulated cycles.
Inhibin B levels in the normal menstrual cycle progressively rise in the early follicular phase and peak around days five to six of the cycle. We have chosen day five or six of the cycle to take the blood sample coinciding with the peak levels. Due to the progressive increase in inhibin B from days one to six of the follicular phase, it was important to take samples on the same day of the cycle so comparisons between patients in the study could be made. We also wanted the samples to be taken near peak concentrations so the measurement of inhibin B would be reliable and precise than measuring levels close to the detection limit of the assay.
Some studies report that inhibin B measurement improves the prediction of ovarian reserve in IVF patients.4–6 Whereas, other studies report that inhibin B does not improve the predictive value of existing markers (age and FSH) of ovarian reserve.7,8 However, this study has shown that in the poor responders, inhibin B on day five of the menstrual cycle is a better marker of ovarian response to IVF treatment than FSH. The discrepancies between the above studies could be due to the variable time (days one to five) at which the samples were analysed by the different groups. Because inhibin B levels are progressively increasing in the early follicular phase, it is important to take the samples on a particular day of the cycle to obtain a reliable relationship between inhibin B and ovarian reserve.
AMH is expressed by granulosa cells of small growing follicles.16,17 In the follicular fluid, AMH concentrations are higher in small antral follicles than in larger more differentiated follicles.17 Human female serum contains a measurable amount of AMH during the menstrual cycle13 and women with PCOS have raised levels of AMH.18 Hormonal regulation of AMH in the ovary remains unclear.19–21 Recent reports suggest AMH levels decline with age22 and AMH is associated with ovarian response in IVF patients.11,23 In this study, we found that AMH had the most significant relationship with the number of eggs collected, compared with FSH or inhibin B in poor responders. AMH was also a better marker in predicting a cancelled cycle compared with FSH or inhibin B in this particular group of patients. Our observations are consistent with other reported studies from a wider group of IVF patients.11,23
In conclusion, it is evident that as a single marker AMH provided the best indication of ovarian response in poor responders. Undetectable levels of AMH suggest poor response to gonadotrophin stimulation in IVF patients. Combining FSH, inhibin B and AMH modestly improved the predictive statistics. It will be interesting to investigate whether this predictive value will be improved in poor responders by combining the measure of antral follicle count with these biochemical markers.