Menstrual Phase and Menopausal Status Classification of Benign Breast Tissue Using Hormone-Regulated Gene Expression and Histomorphology: A Validation Study.

BACKGROUND: The validation of breast cancer risk biomarkers in benign breast samples (BBS) is a long-sought goal, hampered by the fluctuation of gene and protein expression with menstrual phase (MP) and menopausal status (MS). Previously, we identified hormone-related gene expression and histomorphology parameters to classify BBS by MS/MP. We now evaluate both together, to validate our prior results. PATIENTS AND METHODS: BBS were obtained from consenting women (86 premenopausal, 55 postmenopausal) undergoing reduction mammoplasty (RM) or contralateral unaffected breast (CUB) mastectomy. MP/MS was defined using classical criteria for menstrual dates and hormone levels on the day of surgery. BBS gene expression was measured with reverse transcription quantitative polymerase chain reaction (RT-qPCR) for three luteal phase (LP) genes (TNFSF11, DIO2, MYBPC1) and four menopausal genes (PGR, GREB1, TIFF1, CCND1). Premenopausal samples were classified into LP or non-LP, using published histomorphology parameters. Logistic regression and receiver-operator curve analysis was performed to assess area under the curve (AUC) for prediction of MP/MS. RESULTS: In all 131 women, menopausal genes plus age > 50 years predicted true MS [AUC 0.93, 95% confidence interval (CI) 0.89, 0.97]. Among premenopausal women, high TNFSF11 expression distinguished non-LP from LP samples (AUC 0.80, 95% CI 0.70, 0.91); the addition of histomorphology improved the prediction nonsignificantly (AUC 0.87, 95% CI 0.78, 0.96). In premenopausal subsets, addition of histomorphology improved LP prediction in RM (AUC 0.95, 95% CI 0.87, 1.0), but not in CUB (0.84, 95% CI 0.72, 0.96). CONCLUSIONS: Expression of five-gene set accurately predicts menopausal status and menstrual phase in BBS, facilitating the development of breast cancer risk biomarkers using large, archived sample repositories.

Effects of web-based cognitive behavioral stress management and health promotion interventions on neuroendocrine and inflammatory markers in men with advanced prostate cancer: A randomized controlled trial.

Cognitive behavioral stress management (CBSM) improves quality of life and mitigates stress biology in patients with early-stage cancer, including men with localized prostate cancer. However, treatments for advanced prostate cancer like androgen deprivation therapy (ADT) can lead to significant symptom burden that may be further exacerbated by stress-induced inflammation and cortisol dysregulation. The aim of this study was to examine the effects of CBSM (versus an active health promotion control) on circulating inflammatory markers and cortisol in men with advanced prostate cancer. METHODS: Men with stage III or IV prostate cancer (N = 192) who had undergone ADT within the last year were randomized to CBSM or health promotion. Both interventions were 10 weeks, group-based, and delivered online. Venous blood was drawn at baseline, 6 months, and 12 months to measure circulating levels of CRP, IL-6, IL-8, IL-10, and TNF-α. Saliva samples were collected at awakening, 30 min after awakening, evening, and night for two consecutive days at baseline, 6-months, and 12-months to measure diurnal cortisol slopes. RESULTS: Mixed modeling analyses demonstrated that changes in inflammatory markers and cortisol did not differ by intervention. Men in both CBSM and health promotion showed decreases in IL-10, IL-8, and TNF-α from baseline to 6 months (ß = -3.85--5.04, p's = 0.004-<0.001). However, these markers generally demonstrated a rebound increase from 6 to 12 months (ß = 1.91-4.06, p's = 0.06-<0.001). Men in health promotion also demonstrated a flatter diurnal cortisol slope versus men in CBSM at 6 months (ß = -2.27, p = .023), but not at 12 months. There were no intervention effects on CRP, IL-6, or overall cortisol output. CONCLUSIONS: Contrary to hypotheses, CBSM did not lead to changes in the circulating inflammatory markers and cortisol relative to health promotion. CBSM may be associated with healthy diurnal cortisol rhythm because of its focus on cognitive behavioral approaches to stress management. More research is needed to understand the impact of CBSM and health promotion on biomarkers among men with advanced prostate cancer.

Shift from androgen to estrogen action causes abdominal muscle fibrosis, atrophy, and inguinal hernia in a transgenic male mouse model.

Inguinal hernia develops primarily in elderly men, and more than one in four men will undergo inguinal hernia repair during their lifetime. However, the underlying mechanisms behind hernia formation remain unknown. It is known that testosterone and estradiol can regulate skeletal muscle mass. We herein demonstrate that the conversion of testosterone to estradiol by the aromatase enzyme in lower abdominal muscle (LAM) tissue causes intense fibrosis, leading to muscle atrophy and inguinal hernia; an aromatase inhibitor entirely prevents this phenotype. LAM tissue is uniquely sensitive to estradiol because it expresses very high levels of estrogen receptor-α. Estradiol acts via estrogen receptor-α in LAM fibroblasts to activate pathways for proliferation and fibrosis that replaces atrophied myocytes, resulting in hernia formation. This is accompanied by decreased serum testosterone and decreased expression of the androgen receptor target genes in LAM tissue. These findings provide a mechanism for LAM tissue fibrosis and atrophy and suggest potential roles of future nonsurgical and preventive approaches in a subset of elderly men with a predisposition for hernia development.

The relationship of single-strand breaks in DNA to breast cancer risk and to tissue concentrations of oestrogens.

CONTEXT: Clinical study of breast cancer patients in Chicago, IL, USA. OBJECTIVE: Ascertain the utility of measurements of single-strand breaks (SSB) in DNA for assessment of breast cancer risk. METHODS: Fine-needle aspirates of the breast, SSB by nick translation, percent breast density (PBD), Gail model risk, cumulative methylation index (CMI), enzymes of DNA repair and tissue antioxidants. RESULTS: DNA repair enzymes and 4-hydroxyestradiol were negatively associated with SSB; CMI and PBD were positively associated. CONCLUSIONS: Quantitative measurement of SSBs by this procedure indicates the relative number of SSBs and is related to promoter methylation, antioxidant availability and percent breast density.

Genetic Determinants of Nipple Aspiration Fluid Yield.

PURPOSE: Nipple aspiration fluid (NAF) is a non-invasively-acquired biosample that can provide a window into the breast environment, but NAF yield is highly variable. Its determinants must be better understood for studies of breast cancer risk. The wet earwax phenotype was identified as one determinant of NAF yield in the 1970s, and is linked to single nucleotide polymorphisms (SNPs) in the ATP-binding cassette transporter gene ABCC11. We have investigated this, as well as SNPs in the prolactin (PRL) and prolactin receptor (PRLR) genes, in relation to NAF yield. METHODS: DNA was extracted from white blood cells of 557 NAF yielders and 359 non-yielders, and was used to genotype ABCC11 (rs17822931), PRL (rs849870, rs849872, rs849886, rs2244502, rs1341239), and PRLR (rs37364, rs34024951, rs1610218, rs9292575, rs7718468) using Taqman genotyping assay. The association between NAF yield and each single SNP was analyzed using logistic regression adjusting for age, race, and menopausal status. RESULTS: ABCC11 rs17822931 showed a negative association with NAF yield [odds ratio (OR) 0.66, 95 % confidence interval (CI) 0.49-0.88; p = 0.004]. The PRL rs849870 and the haplotype combination with other SNPs showed a marginal association with NAF yield. In addition, the years since last birth also showed negative association with NAF yielding (OR 0.98, 95 % CI 0.96-0.99; p = 0.001). The combination of the years since last birth with ABCC11 SNP revealed significant interaction between reproductive factor and genetic factor. CONCLUSIONS: Our results confirmed the association between NAF yield and earwax phenotype through ABCC11 genotype. Combined with the recency of last birth, ABCC11 genotype should be considered in the design of studies utilizing NAF as a biosample.

RANKL expression in normal and malignant breast tissue responds to progesterone and is up-regulated during the luteal phase.

The receptor activator of nuclear factor-κB ligand (RANKL) acts as a paracrine factor in progesterone-induced mammary epithelial proliferation and tumorigenesis. This evidence comes mainly from mouse models. Our aim was to examine whether RANKL expression in human normal and malignant breast is under the control of progesterone throughout the menstrual cycle. Breast epithelial samples were obtained by random fine needle aspiration (rFNA) of the contralateral unaffected breasts (CUB) of 18 breast cancer patients, with simultaneous serum hormone measurements. Genes correlated with serum progesterone levels were identified through Illumina microarray analysis. Validation was performed using qRT-PCR in rFNA samples from CUB of an additional 53 women and using immunohistochemistry in tissue microarrays of 61 breast cancer samples. Expression of RANKL, DIO2, and MYBPC1 was correlated with serum progesterone in CUB, and was significantly higher in luteal phase. RANKL and MYBPC1 mRNA expression were highly correlated between CUB and matched tumor samples. RANKL protein expression was also significantly increased in the luteal phase and highly correlated with serum progesterone levels in cancer samples, especially in hormone receptor positive tumors. The regulatory effects of progesterone on the expression of RANKL, DIO2, and MYBPC1 were confirmed in three-dimensional cultures of normal breast organoids. In normal breast and in breast cancer, RANKL mRNA and protein expression fluctuate with serum progesterone with highest levels in the luteal phase, suggesting that RANKL is a modulator of progesterone signaling in normal and malignant breast tissue and a potential biomarker of progesterone action and blockade.

Relationship of Estradiol and Progesterone with Partnership and Parity Among Bangladeshi and British Women of European Origin.

Recent studies in social endocrinology have explored the effects of social relationships on female reproductive steroid hormones-estradiol and progesterone-investigating whether they are suppressed in partnered and parous women. Results have been mixed for these hormones although evidence is more consistent that partnered women and women with young children have lower levels of testosterone. These studies were sequential to earlier research on men, based on Wingfield's Challenge Hypothesis, which showed that men in committed relationships, or with young children, have lower levels of testosterone than unpartnered men or men with older or no children. The study described here explored associations between estradiol and progesterone with partnership and parity among women from two different ethnicities: South Asian and white British. We hypothesized that both steroid hormones would be lower among partnered and/or parous women with children ≤3 years old, regardless of ethnicity. In this study we analyzed data from 320 Bangladeshi and British women of European origin aged 18 to 50 who participated in two previous studies of reproductive ecology and health. Levels of estradiol and progesterone were assayed using saliva and/or serum samples and the body mass index calculated from anthropometric data. Questionnaires provided other covariates. Multiple linear regressions were used to analyze the data. The hypotheses were not supported. We argue here that, unlike links between testosterone and male social relationships, theoretical foundations for such relationships with female reproductive steroid hormones are lacking, especially given the primary role of these steroids in regulating female reproductive function. Further longitudinal studies are needed to explore the bases of independent relationships between social factors and female reproductive steroid hormones.

Prediction of menopausal status from estrogen-related gene expression in benign breast tissue.

The utility of archived paraffin-embedded breast tissue for risk-related research is often limited by missing menopausal status data. We tested the hypothesis that breast tissue gene expression patterns can improve menopausal stratification. Healthy high-risk participants in a clinical trial underwent breast random fine-needle aspiration (rFNA); 100 ng of RNA extracted from rFNA samples was reverse-transcribed; the expression of 28 estrogen-responsive genes was evaluated by real-time PCR. True menopausal status (TMS) was determined by measurement of plasma hormones and age. Differentially expressed genes and age were analyzed by logistic regression. The accuracy of the menopause prediction was assessed using receiver-operator characteristic (ROC) analysis, and validated in a second independent set of 44 women. In the test set, postmenopausal women demonstrated significantly lower expression of five estrogen-responsive genes: GREB1, PGR, TFF1, PRLR, and CCND1 (adjusted P < 0.03 for all). In the validation set, three of these genes were expressed at lower levels in postmenopausal women (GREB1, PGR, TFF1) (adjusted P < 0.06 for all). In the test set, the modeled area under the curve (AUC) for age and three genes was higher than for age >50 alone (AUC 96.1% vs. 87.2%, P = 0.002), and remained better than for age alone in the validation set (99.0% vs. 95.5%, P = 0.16). Estrogen-related gene expression in breast specimens can be used to improve menopausal classification, reducing the biological noise related to menopause in studies that seek to identify RNA or protein risk biomarkers in archived breast samples.

Functions of dehydroepiandrosterone in relation to breast cancer.

Although DHEA sulfate (DS) is the most abundant steroid in the circulation, breast fluid contains an approximately 80-fold greater concentration than serum. Transport of DS into cells requires organic anion transporting polypeptides (OATPs), which are specific for cell type, cell location, and substrate, but may have a broader specificity for housekeeping functions. Specific classes, which may be modified by soluble factors including neutral steroids, have been identified in the breast. After transport, DS may be cleaved to DHEA by ubiquitous sulfatases, which may be modified by the cell milieu, or DHEA may enter by diffusion. Synthesis from cholesterol does not occur because CYP17B12 and cytochrome b5 are lacking in breast tissues. Case-control studies reveal a positive association of serum DS with risk of breast cancer. The association is even greater with DHEA, particularly in postmenopausal women with HR + invasive tumors. Metabolites of DHEA, androstenedione and testosterone, are associated with breast cancer but DHEA is likely to have an independent role as well. Mechanisms by which DHEA may promote breast cancer relate to its effect in increasing circulating IGF-I, by inhibiting the suppressive effect of glucocorticoids, and by promoting retention of pre-adipocytes with aromatase activity. In addition, DHEA may interact with the G-protein coupled receptor GPER for stimulation of miR-21 and subsequent activation of the MAPK pathway. DHEA also has antitumor properties that relate to stimulation of immunity, suppression of inflammation, and elevation of adipose tissue adiponectin synthesis. The net effect may depend on the which factors predominate.

HPLC fractionation with immunoassay of steroids from nipple aspirate fluid.

Fractionation of steroids allows for multiple assays to be run on a single low volume liquid biopsy, whereas performing the same number of assays without fractionation would require increasing the sample volume by dilution, rendering the concentration of steroids below the level of detection for most, if not all, downstream assays. Briefly, steroids are extracted from a biofluid sample using solvent phase extraction to separate the aqueous (conjugated) steroids from the non-aqueous (non-conjugated) steroids in the organic phase. The latter is further separated by high-performance liquid chromatography (HPLC) and collected in an automated fraction collector based on the UV detection of internal standards. Commercially available immunoassays are then used to quantify the < ng/ml concentrations of steroids in each fraction. This protocol was designed for small samples of nipple aspirate fluid (minimum 2 µL), but it can be modified to fractionate steroids from homogenized solid tissue samples or other liquid biopsies. Included in this protocol are precautions to help ensure reproducibility and minimize matrix effects and other errors of measurement, given that samples requiring fractionation are fundamentally precious and, like other quantitative procedures of small samples, can be prone to contamination by solvent residues and other factors.•The method permits quantitative analysis of multiple steroids from very small volumes of biofluid.•Fractionation by HPLC provides a highly purified sample for quantification.•The immunoassay end point provides specificity without expensive equipment.

Association of genetic polymorphisms with local steroid metabolism in human benign breasts.

PURPOSE: Although alterations of concentrations in circulating steroids have been linked to single nucleotide polymorphisms (SNPs) of steroidogenic enzymes, we hypothesized that SNPs of such enzymes located within the breast affect local steroid concentrations more than products of such SNPs absorbed from the circulation. METHODS: Steroids (estradiol, estrone, testosterone, androstenedione, DHEA, DHEA sulfate, progesterone) in nipple aspirate fluid (NAF) were purified by HPLC and they along with serum steroids were quantified by immunoassays. Polymorphisms of the transporter SLCO2B1 and enzymes HSD3B1, CYP19A1, HSD17B12, AKR1C3, CYP1B1, and SRD5A1 were measured in white blood cell DNA. RESULTS: Steroid concentrations in NAF of subjects with homozygous minor genotypes differed from those with heterozygotes, i.e., SLCO2B1 (rs2851069) decreased DHEAS (p = 0.04), HSD17B12 (rs11555762) increased estradiol (p < 0.004), and CYP1B1 (rs1056836) decreased estradiol (p = 0.017) and increased progesterone (p = 0.05). Also, in serum, CYP19A1 (rs10046 and rs700518) both decreased testosterone (p = 0.02) and SRD5A1 increased androstenedione (p = 0.006). Steroids in subjects with major homozygotes did not differ from those with heterozygotes indicating recessive characteristics. CONCLUSIONS: In the breast, SNPs were associated with decreased uptake of DHEAS (SLCO2B1), increased estradiol concentrations through increased oxidoreductase activity (HSD17B12), or decreased estradiol concentrations by presumed formation of 4-hydroxyestradiol (CYP1B1). CYP19A1 was associated with decreased testosterone concentrations in serum but had no significant effect on estrogen or androgen concentrations within the breast. The hormone differences observed in NAF were not usually evident in serum, indicating the importance of assessing the effect of these SNPs within the breast.

A prospective study of variability in mammographic density during the menstrual cycle.

Mammographic breast density has been proposed as a surrogate endpoint in breast cancer prevention studies, but little is known about its variability over time, particularly in relation to menstrual cycle phase. The purpose of this study was to assess variation in breast density on digital mammograms using quantitative and qualitative density measures. Menstrual cycle phase was determined by salivary estradiol and progesterone assays. 73 healthy subjects with regular menses had 1-3 mammograms with paired saliva collection during a 12-month period. The mean difference in density as a percentage of the mean density was calculated for follicular-luteal (n = 50), luteal-luteal (n = 26) and follicular-follicular (n = 23) pairs in the same woman using the same breast. Two density measures (measurement of dense area and BIRADS) were used. The mean luteal density exceeded the mean follicular density by 7.1-9.2%, but density differences between luteal pairs and follicular pairs did not exceed 5%. The intraclass correlation for measurement of dense area was greater than 85% in all phases of the menstrual cycle, but was below 50% for BIRADS for luteal-follicular and follicular-follicular pairs. Our study provides estimates of the amount of variation in mammographic density during the menstrual cycle, and that inherent in repeated density measurement in premenopausal women, and suggests that menstrual phase of mammographic evaluation should be controlled for in intervention studies where density is being used as a surrogate measure.

Association of serum prolactin and oxytocin with milk production in mothers of preterm and term infants.

The present study was designed to compare milk production and hormone responses (prolactin [PRL], oxytocin [OT]) and to determine associations of hormone levels with milk production in mothers of preterm (PT) and term (TM) infants during the first 6 weeks postpartum. Mothers of PT infants (n = 95) were all pump dependent; mothers of TM infants (n = 98) were all feeding their infant at breast. Mothers of nonnursing PT infants produced less milk over time compared to mothers of TM infants. A higher proportion of PT mothers had lower basal PRL levels compared with TM mothers. PRL and frequency of breast stimulation combined positively influenced milk production in PT mothers. OT levels were higher in PT versus TM mothers, but OT was not related to milk production. Further study is warranted regarding interventions to enhance milk production, particularly in pump-dependent mothers of PT infants.

Author Correction: Childhood ecology influences salivary testosterone, pubertal age and stature of Bangladeshi UK migrant men.

In the version of this Article originally published, the units for the 'Weight' column in Table 1 were incorrect; they should have been kg. This has now been corrected.

Childhood ecology influences salivary testosterone, pubertal age and stature of Bangladeshi UK migrant men.

Male reproductive investment is energetically costly, and measures of human reproductive steroid hormones (testosterone), developmental tempo (pubertal timing) and growth (stature) correlate with local ecologies at the population level. It is unclear whether male reproductive investment in later life is 'set' during childhood development, mediated through adulthood, or varies by ethnicity. Applying a life-course model to Bangladeshi migrants to the United Kingdom, here we investigate plasticity in human male reproductive function resulting from childhood developmental conditions. We hypothesized that childhood ecology shapes adult trade-offs between reproductive investment and/or other fitness-related traits. We predicted correspondence between these traits and developmental timing of exposure to ecological constraints (Bangladesh) or conditions of surplus (United Kingdom). We compared: Bangladesh sedentees (n = 107); Bangladeshi men who migrated in childhood to the United Kingdom (n = 59); migrants who arrived in adulthood (n = 75); second-generation UK-born and raised children of Bangladeshi migrants (n = 56); and UK-born ethnic Europeans (n = 62). Migration before puberty predicted higher testosterone and an earlier recalled pubertal age compared with Bangladeshi sedentees or adult migrants, with more pronounced differences in men who arrived before the age of eight. Second-generation Bangladeshis were taller, with higher testosterone than sedentees and adult migrants, and higher waking testosterone than Europeans. Age-related testosterone profiles varied by group, declining in UK migrants, increasing in sedentees, and having no significant relationship within UK-born groups. We conclude that male reproductive function apparently remains plastic late into childhood, is independent of Bengali or European ethnicity, and shapes physiological trade-offs later in life.

Breast Hormone Concentrations in Random Fine-Needle Aspirates of Healthy Women Associate with Cytological Atypia and Gene Methylation.

Sex steroid hormones contribute to breast cancer development, but data on concentrations of these within breast tissue are limited. We performed simultaneous multiparameter measurement of breast sex steroids, breast epithelial cytology, and DNA methylation in 119 healthy women (54 pre- and 65 postmenopausal) without a history of breast cancer. Random fine-needle aspiration (rFNA) of the breast was performed simultaneously with blood collection. Breast samples were analyzed by LC/MS-MS for estrone, estradiol, progesterone, androstenedione, and testosterone. Blood samples were assayed for estradiol and progesterone by immunoassay. Cytomorphology was classified using the Masood Score, and DNA methylation of eight genes was analyzed using quantitative multiplexed methylation-specific PCR, and expressed as the cumulative methylation index (CMI). Serum and breast concentrations of estradiol and progesterone showed significant correlation (Spearman r = 0.34, Padj = 0.001 and r = 0.69, Padj < 0.0006, respectively). Progesterone concentration was significantly higher in the premenopausal breast (Padj < 0.0008), and showed a luteal surge. Breast estrone and estradiol concentrations did not differ significantly by menopause, but androstenedione concentration was higher in the breasts of postmenopausal women (P = 0.026 and Padj = 0.208). Breast androgens were significantly correlated with breast density (Spearman r = 0.27, Padj = 0.02 for testosterone) and CMI (Spearman r = 0.3, Padj = 0.038 for androstenedione). Our data indicate that future larger studies of breast steroid hormones along with other parameters are feasible. Significant associations of breast androgen concentrations with breast density and gene methylation warrant future study. Cancer Prev Res; 11(9); 557-68. ©2018 AACR.

Childhood conditions influence adult progesterone levels.

BACKGROUND: Average profiles of salivary progesterone in women vary significantly at the inter- and intrapopulation level as a function of age and acute energetic conditions related to energy intake, energy expenditure, or a combination of both. In addition to acute stressors, baseline progesterone levels differ among populations. The causes of such chronic differences are not well understood, but it has been hypothesised that they may result from varying tempos of growth and maturation and, by implication, from diverse environmental conditions encountered during childhood and adolescence. METHODS AND FINDINGS: To test this hypothesis, we conducted a migrant study among first- and second-generation Bangladeshi women aged 19-39 who migrated to London, UK at different points in the life-course, women still resident in Bangladesh, and women of European descent living in neighbourhoods similar to those of the migrants in London (total n = 227). Data collected included saliva samples for radioimmunoassay of progesterone, anthropometrics, and information from questionnaires on diet, lifestyle, and health. Results from multiple linear regression, controlled for anthropometric and reproductive variables, show that women who spend their childhood in conditions of low energy expenditure, stable energy intake, good sanitation, low immune challenges, and good health care in the UK have up to 103% higher levels of salivary progesterone and an earlier maturation than women who develop in less optimal conditions in Sylhet, Bangladesh (F9,178 = 5.05, p < 0.001, standard error of the mean = 0.32; adjusted R(2) = 0.16). Our results point to the period prior to puberty as a sensitive phase when changes in environmental conditions positively impact developmental tempos such as menarcheal age (F2,81 = 3.21, p = 0.03) and patterns of ovarian function as measured using salivary progesterone (F2,81 = 3.14, p = 0.04). CONCLUSIONS: This research demonstrates that human females use an extended period of the life cycle prior to reproductive maturation to monitor their environment and to modulate reproductive steroid levels in accordance with projected conditions they might encounter as adults. Given the prolonged investment of human pregnancy and lactation, such plasticity (extending beyond any intrauterine programming) enables a more flexible and finely tuned adjustment to the potential constraints or opportunities of the later adult environment. This research is the first, to our knowledge, to demonstrate a postuterine developmental component to variation in reproductive steroid levels in women.

Predictors of preterm infant feeding methods and perceived insufficient milk supply at week 12 postpartum.

As part of a large, nonexperimental, prospective, longitudinal study, 94 pump-dependent mothers of a nonnursing preterm infant were queried via telephone at weeks 8 to 12 post-partum about their infant feeding method. At week 12, 44.6% provided own mother's milk, 26.6% provided own mother's milk + artificial milk, and 28.7% provided artificial milk only. Logistic regression analyses identified the following predictors for risk of artificial milk at week 12 postpartum: multiple birth, week 6 inadequate milk supply, maternal age younger than 29 years, and intended length of lactation less than 34 weeks. Predictors for risk of maternal perceived insufficient milk supply for weeks 8 to 12 postpartum included week 6 inadequate milk supply, unemployment, and infant hospital discharge after postpartum day 42. Further research is needed to assist pump-dependent mothers of preterm infants with sustaining their milk supply.