Aron Johnson, Phillip Patton, Susan Titus, David Battaglia, and Kenneth Burry

Department of Obstetrics and Gynecology, Oregon Health Sciences University, Portland, OR.

Forward:
    Fertility clinics across the country commonly use patient follicle stimulating hormone (FSH) values as a means of predicting a patient’s potential response to fertility treatments. As patients age, their FSH values tend to increase, particularly as they near the perimenopause period.  Fertility begins to decline with increasing age, so lower FSH values are usually indicators of a better response to treatment than are higher values. The challenge for the reproductive endocrinologist is to determine a reasonable FSH cutoff value that can select for patients with a reasonable chance of success. Above this value the patient would be best advised to pursue other fertility options (e.g. donor oocytes).
    OHSU Fertility Consultants has investigated the relationship between FSH level and fertility success in IVF.  Accordingly, we have established a relatively high FSH cut-off value ( >15) as compared to other fertility programs where FSH cut-off values are considerably lower (>10).  Because of our successes with patients with elevated FSH values, we have set this higher cut-off to prevent the elimination of patients who still have a reasonable chance of success with IVF.  Thus, we will accept a patient into our program with an FSH of 14 when other clinics would have excluded them.
    We have performed a study to determine if our FSH cut off value was appropriate and clinically justified.  By examining FSH values and  pregnancy outcome over a 2 year period we found that no one achieved a pregnancy with an FSH level above 15. However, 17 % of our total pregnancies occurred in patients with FSH values between10 to 15!  Thus, if we had adopted the more traditional FSH cut off value of 10 (the value that many other clinics use) a significant number of our patients would been denied their dream of having children using their own eggs.
The philosophy behind why some clinics have a lower FSH cut offs is to eliminate couples that may have a lower chance of becoming parents using their own eggs. By only allowing couples into their ivf program that have a better chance of pregnancy they attempt to generate a comparatively higher ivf success rate. However, our experience shows that many of those patients, who would be considered to have a poor prognosis in other programs, certainly did well in our program.
    Our study also demonstrated how reproducible and similar FSH values are between different fertility programs. We found that the FSH values generated in our laboratory are very similar to FSH values obtained in other fertility laboratories. Therefore, it is easy to compare our values with those generated at other programs.  We believe that our higher FSH cut-off could  also be used effectively at other clinics.
    Because FSH levels are most often related to a patient’s age, FSH determination is usually performed on older patients ( >35 years old). Since age is the most important factor influencing female fertility, the overall pregnancy rates described here represents an older population that, as a group, experience a lower pregnancy rate compared to our younger patients. Please see our overall results to examine our last years performance in all patient groups.

Background and Objective:

In the Clomiphene Citrate Challenge Test, (CCCT) FSH values on days 3 and 10 are used to calculate ovarian reserve and direct treatment. However, there is little agreement about threshold FSH values, (range 10 to 15 mIU/ml) to achieve these objectives.

In a program that uses a high threshold FSH, (maxFSH > 15 mIU/ml), we analyzed FSH values and pregnancy outcomes in women participating in IVF. Our objective was to define an FSH threshold that would identify patients with a poor prognosis for pregnancy but without significantly limiting the number of pregnancies.

Population and Analysis:

All women > 35 years of age complete a standard CCCT prior to IVF. In a retrospective review of patients' cumulative IVF cycle outcomes for both fresh and frozen embryo transfers were analyzed from 1/1/2001 to 3/1/2003 where a maxFSH value was available. Donor egg cycles were excluded. For individuals experiencing multiple trials but failed to ever become pregnant we considered their age at the time first transfer. When multiple pregnancy cycles occurred, only the first cycle was included.

We then performed a stepwise analysis of pregnancy outcome as a function of maxFSH cutoff values that ranged from 7 to 15 mIU/ml.

Methods:

FSH testing was performed by a commercially available automated chemiluminescent immuno-assay (IMMULITE ® Diagnostic Products Inc.). The system uses a solid phase antibody specific coated plastic bead with an alkaline phosphatase labeled reagent and chemiluminescent substrate.

Results:

Of 350 individuals 113 experienced a pregnancy (32.3 %). The mean age was 39.1 years ranging from 35 to 47 years, (fig 1). No pregnancies occurred in the 11 cases having a maxFSH >15. There was no clinically significant difference in the overall success rate using a maxFSH >15 Vs >10, 33.3 % and 34.6 % respectively, (fig 2). However, 19 of the potential 113 pregnancies, (17%) would have been excluded by a maxFSH cut off > 10, (fig 3 & fig 4). Using lower maxFSH cut off values that begin to significantly enhance pregnancy rates occur only at a cost of excluding over a fifth of the pregnancies, (fig 2, maxFSH > 9, 36.3% related to a maxFSH > 9, 79 % of total pregnancies).

Quality Control and Inter-Laboratory Relevance:

Three factors are important to consider when evaluating the clinical significance of data from a longitudinal study to that of other centers managing fertility:

1. inter-assay variation; How do the same control source FSH values relate over a significant period of time when performed by a single laboratory?

2. inter-laboratory variation; How do the same source FSH values relate among different laboratories?

3. analyzer confounding; How do the same source FSH values relate when performed on analyzers of different manufactures?

Inter-assay variation is determined by testing the same controls with every assay run. All reported FSH values over the period of this study were from test runs in which the control values met the laboratory criteria for acceptability. The same lot of controls tested for182 consecutive runs at 3 values (low, medium, and high) resulted in means of: 5.1 ± 0.17 sd = 3.3% c.v.; 16.9 ± 0.84 sd = 5.0% c.v. and 44.2 ± 2.1 sd = 4.7 c.v. Of the hormone assays, FSH is clearly the most stable and reproducible. We have never had to exclude an FSH assay run due to the controls being out of acceptable limits.

The degree of analyzer confounding and inter-laboratory variation can both be obtained from our participation in quarterly proficiency testing services provided by the American Association of Bioanalysts (AAB). Laboratories perform quarterly FSH testing on 3 common broadcast specimens. Results are published by the AAB and inter-laboratory performance can be obtained. By example the independent mean FSH value on the same specimen for the 305 reporting laboratories for the 3rd quarter 2003 was 12.9 ± 1.4 sd 10.9% c.v.. Figure 5 shows the portion of laboratories using the same analytical system that acquired these data. We use the most popular system similar to 1/3 of all reporting laboratories. Figure 6 shows the mean FSH results of the 305 laboratories using the same analytical system. Two of the most unpopular 7th and 9th systems added the largest amount of confounding. Conversely, the difference of the means for the 6 most popular laboratory systems representing 90 % of all laboratories, (mean = 13.0 mIU/ml) varied by an average of 0.7 (range 0.3 to 1.2) or 5%.

Discussion:

FSH is a relative predictor of IVF outcome. FSH levels of 10 to 12 mIU/ml have been used to discriminate participation in IVF. This threshold may exclude a significant number of couples since we have shown pregnancies up to 15 mIU/ml. Clinics should consider their patients' wishes over their programs published pregnancy rates.

Conclusions:
1)   Many programs exclude IVF candidates based on baseline or stimulated FSH values that exceed an arbitrary and possibly artificial value (generally FSH values > 10 mIU/ml). However, one sixth of pregnancies, 17%, in our 35 to 47 year old patient group were achieved with patients having FSH values > 10 mIU/ml.

3)   Despite the suspicion that FSH assays vary significantly among laboratories, the AAB survey indicates that the majority of reporting programs use one of three assay systems. Furthermore, FSH values reported from various laboratories using different assay systems are remarkably similar and allows for an equitable and more confident comparison of FSH values among the majority of fertility centers.