Study design and population
This prospective longitudinal study was approved by our institutional review board and written informed consent was obtained from all participants.
A total of 40 healthy fertile volunteers without family history of breast or ovarian cancer entered this study. All women underwent initial clinical examination and breast ultrasound to check for compliance with inclusion criteria. Exclusion criteria were: previous breast surgery, recent breast feeding, presence of breast nodules or cysts larger than 1 cm, and contraindications to undergo MRI.
Volunteers were divided into two groups according to whether they had a physiological menstrual cycle or they used OC, as follows: a group called “nOC” included 20 women with physiological menstrual cycle, aged 28 ± 3 years (mean ± standard deviation; median 27); a group called “OC” included 20 women using OC, aged 26 ± 3 years (median 26). The mean body mass index ± standard deviation was 21 ± 3 kg/m2 for the nOC group and 20 ± 2 kg/m2 for the OC group; the mean age at menarche ± standard deviation was 12 ± 1 year and 13 ± 1 year, respectively.
Hormonal assay
Each participant underwent a hormonal assay using a radioimmunoassay technique based on labeled antigen–antibody binding [21]. Estradiol, progesterone, follicle stimulating hormone, luteinising hormone, and prolactin were measured. Women in the nOC group underwent hormonal assay three times, on days 7, 14, and 21 of the menstrual cycle; women in the OC group underwent hormonal assay only once, on day 14.
1H-MRS protocol
As for the hormonal assay, women in the nOC group underwent 1H MRS three times, on days 7, 14, and 21 of the menstrual cycle; women in the OC group underwent 1H MRS only once, on day 14.
All MRS studies were performed using a 3-T unit (General Electric Healthcare, Discovery MR 750, Little Chalfont, UK). Automated shimming was performed before each examination. Water suppressed single-voxel MRS was acquired using a point-resolved MRS sequence (repetition time 1,500 ms, echo time 135 ms, number of excitations 82) and a bilateral dedicated eight-channel breast coil with the patient in the prone position. A repetition time of 1,500 ms was chosen to shorten the acquisition time and avoid participants’ movement. An isotropic voxel of 2 × 2 × 2 cm3 was placed on the most homogeneous part of the mammary gland, thus avoiding partial-volume effects (Fig. 1).
All spectra were processed by a physicist with 5 years of experience using the software provided by the manufacturer (SAGE, v07, General Electric Healthcare, Little Chalfont, UK). After filtering and baseline and phase correction, the peak amplitude of tCho (centred at about 3.2 ppm) was measured in arbitrary units (au).
The formula by Bolan [22] was used to correct the peak amplitude of tCho (tCho'):
$$ {tCho}^{\hbox{'}}=\frac{tCho}{f_{gain}{f}_{coil}{f}_{T_1}{f}_{T_2}} $$
with correction factors:
$$ {f}_{gain}= gain/{gain}_0 $$
$$ {f}_{coil}={B}_1/{B}_{1,0} $$
$$ {f}_{T_1}\approx 1-{e}^{-\frac{TR}{T_1}} $$
$$ {f}_{T_2}={e}^{-\frac{TE}{T_2}} $$
where gain is the receiver gain and B1 is the local amplitude of the excitation radiofrequency field, both retrieved from DICOM metadata. T1 and T2 relaxation times of choline were taken from the literature as being 870 ms and 400 ms, respectively. After these corrections are made, the signal amplitude is proportional to the number of nuclei in the volume [22].
To quantify the absolute [tCho], expressed as mM, the same MRS protocol and the same fitting procedure were applied once to a commercial phantom containing choline at a known concentration of 2 mM. In general, the corrected signal amplitude A' of a resonance is proportional to the number of nuclei n in the sample: n = ksysA'. The system constant ksys accounts for the system-specific hardware and software. The externally referenced [tCho] was then expressed as:
$$ \left[\mathrm{tCho}\right]=\frac{tCho}{f_{gain}{f}_{coil}{f}_{T_1}{f}_{T_2}}\frac{k_{sys}}{\rho_{ph}} $$
where ρph was the phantom density, assumed to be 1 kg/L.
Statistical analysis
The statistical analysis was performed using the SPSS software package (SPSS IBM inc. v19, Chicago, IL, USA). A p value < 0.05 was regarded as significant.
Considering the small sample size, continuous tCho data were reported as median and interquartile interval (25th–75th percentile) and non-parametric statistics were adopted.
Age, body mass index, and age at menarche were presented as mean ± standard deviation; differences between the two groups were ascertained using the Student t test.
For the nOC group, changes over the menstrual cycle of [tCho] and hormones were assessed using the Friedman test. The [tCho] measured in the OC group was compared to [tCho] measured in the nOC group at days 7, 14, and 21 of the menstrual cycle using the Mann-Whitney U test. In both groups, [tCho] was correlated to concentration of hormones using the Spearman correlation coefficient.