SHIPPING ABOVE €200
SinglePheromones.com
e-mail: [email protected]
e-mail: [email protected]
BIG SUMMER SALE! UP TO -25%!
days,
hours,
minutes,
seconds.
BIG SUMMER SALE! UP TO -25%!
BACKGROUND: To date, there has not been an investigation to determine whether lactating women and their infants influence the ovarian function of other women with whom they interact. METHODS: In a randomized, double‐blind, placebo‐controlled study of 47 nulliparous women, we utilized both within‐ and between‐subjects controls to assess the effects of sustained exposure to breastfeeding compounds on menstrual cycle length, as well as characteristics of each phase of the ovarian cycle. RESULTS: Breastfeeding compounds modulated ovarian cycle length in comparison with the carrier control (0.01 ≤ all P values ≤ 0.05), disrupting the normal homeostatic regulation of cycle length and tripling its variance. Hence, women with long cycles stayed long and did not regress to the mode of 29 days and women with short cycles maintained short cycles. This effect was driven by changes in both the follicular and luteal phases of the cycle (0.01 ≤ all P values ≤ 0.04) and changed the timing of the pre‐ovulatory surge of LH. CONCLUSIONS: Because compounds from lactating women and their infants modulated the ovarian cycles of women, as is seen in other mammals, they have the potential to function as pheromones, regulating fertility within groups of women.
Pheromones are defined as those natural compounds produced by one member of a social group that can regulate the neuroendocrine mechanisms underlying fertility, development or behavior of another group member. Effective in minute quantities, they are specialized types of social chemosignals used for communication within a species ( Karlson and Butenandt, 1959 ; Karlson and Luscher, 1959 ; McClintock, 2002 ). Initially discovered in insects, there has been great interest in discerning whether humans also utilize social chemosignals as pheromones ( Wysocki and Preti, 1998 ; McClintock, 2000 ).
To date, a few studies have shown that natural compounds produced in the axillary region of women and men can trigger neuroendocrine responses in recipient women and thus are considered to be pheromones. For example, axillary compounds collected from women during the ovulatory phase of their menstrual cycle lengthened the recipient woman’s ovarian cycle, delayed the pre‐ovulatory surge of LH and decreased LH pulsing, whereas those collected during the follicular phase shortened cycle length by accelerating the pre‐ovulatory LH surge and increasing LH pulse frequency ( Stern and McClintock, 1998 ; McClintock, 2000 ; Shinohara et al ., 2001 ). Shortening of the cycle ( Cutler et al ., 1986 ) and alterations in the pulsatile release of LH ( Preti et al ., 2003 ) were also observed when women were exposed to axillary compounds collected from men.
In spite of the fact that females of most mammalian species spend a greater portion of their reproductive life spans in birth cycles of conception, pregnancy, and lactation than in spontaneous unfertilized ovarian cycles ( Altmann et al ., 1978 ; Gudermuth et al ., 1984 ; Hedricks and McClintock, 1985 ; Ellison, 2001 ), there has been little investigation on whether lactating women and their infants influence the ovarian function and behaviour of other women with whom they interact. Animal model studies have revealed that pheromones from lactating rats and their pups induce maternal behaviours in adult conspecifics ( Mennella and Moltz, 1988a ,b, 1989 ) and increase the variability of ovarian cycles in recipient females by lengthening the cycle ( McClintock, 1984 ; Mennella and Moltz, 1989 )
Because prior studies have demonstrated effects of ovarian pheromones in both humans and rats ( McClintock, 1984 ; Cutler, 1987 ; Cutler and Stine, 1988 ; Mennella and Moltz, 1989 ; Stern and McClintock, 1998 ; Shinohara et al ., 2001 ), the present study tested the hypothesis that exposure to chemosignals produced by lactating women and their infants would disrupt ovarian cyclicity, changing the follicular and luteal phases. Specifically, this novel study hypothesized that pheromones from breastfeeding (non‐ovulating) women would increase the variability of ovarian cycles, particularly by lengthening them, and also by shortening them, as the effects of pheromones depend on the state of the ovary at the time of pheromone exposure ( Schank and McClintock, 1992 ). Indeed, phase response curves are an integral part of dynamically complex oscillating systems. Thus, we chose regression analyses that would reveal how breastfeeding compounds might perturb cycle length, based on the ovarian cycles’ initial state just prior to exposure. We also assessed whether sustained exposure to breastfeeding compounds continued to have similar effects in subsequent cycles.
To this aim, we collected natural compounds produced by both members of the breastfeeding dyad—the lactating mother and her infant—and determined the effects of initial and sustained exposure to such compounds on the length of ovarian cycle and its variance in recipient, nulliparous women.
The present study utilized a double‐blind, between‐ and within‐subjects controlled experimental design. Nulliparous women were studied for three consecutive menstrual cycles. During the first cycle, hereafter referred to as the baseline cycle, each woman was exposed to pads moistened with the carrier control of potassium phosphate‐buffered solution (control condition). During the two subsequent experimental cycles, half of the women continued to receive pads with potassium phosphate (control group; n = 30) whereas the remaining women (experimental group; n = 27) were exposed to pads worn next to the axillae and breasts of lactating women, hereafter referred to as breastfeeding compounds. Because each mother nursed her infant several times during the collection periods, the breast secretions most likely also contain secretions from the infant (e.g. saliva). The following sections describe the study design and methods used to collect axillary and breast secretions and to expose other women to these breastfeeding compounds.
Axillary and breast secretions were collected from 26 lactating women (31% African American, 69% Caucasian) living in the Philadelphia area (aged 32.2 ± 1.4 years), who were feeding their infants exclusively by nursing (13 girls, 13 boys; mean age = 3.5 ± 0.3 months) and had not yet resumed menstruation. Each donor participated in the study for 5 or 10 days. All procedures were approved by the Office of Regulatory Affairs at the University of Pennsylvania and informed consent was obtained from each subject before study participation.
Because previous research has shown that dietary flavors can be transmitted to human milk ( Mennella, 1995 ), the mothers were instructed to eat a ‘bland’ diet low in sulphur‐containing foods. To encourage compliance, they were given a list of foods and spices to avoid (e.g. garlic, onion, curry) and were asked to record all foods and beverages consumed during this period. None of the mothers reported consuming such foods or spices during the collection period.
Each day of the collection period, the donors bathed without perfumed products and wore 4×6 inch cotton Webril™ pads (The Kendall Company, USA) in their nursing brassiere for ≥8 h, after which the pads were placed in glass vials [following modified methods used by Russell et al . (1980 ) and Stern and McClintock (1998 )]. Dress shields (J.C.Penny, USA) were provided to secure the axillary pads in place. Nylon gloves (Scientific Instrument Services, USA) were worn by the women and experimenters when handling the pads.
Control pads with carrier solution were treated to match the axillary and breast pads both in moisture and appearance. Potassium phosphate (K 2 HPO 4 ) buffer solution was used as the control carrier solution since the components of this fluid are similar in kind, concentration and pH to that of female sweat and breast milk ( ICRP, 1975 ; Lentner, 1981 ). Each breastfeeding and control pad was cut into four sections for future distribution and then frozen immediately at –80°C in glass vials.
At least 40 women (20 in each group) were required to have an 80% chance of detecting an effect on menstrual cycle length; this is based on the reported SD of 2.5 days for women aged 18–35 years (an alpha of 0.05 and three independent variables). Hence, from January 1998 to March 1999, women were recruited for a 3 month study through posters, newspapers and fliers in a university community in Chicago. Each woman was screened and then included into the study if she was non‐smoking, nulliparous, aged 18–35 years, had a history of regular menstrual cycles, and was not using birth control pills or an intrauterine device. She was included if she was within 30% of ideal body mass index (BMI; 21.5 kg/m 2 ), did not have a history of sinus problems, frequent colds, allergy symptoms, or nasal congestion, did not report psychiatric symptoms, was within normal range of olfaction as assessed by the University of Pennsylvania Smell Identification Test (UPSIT; Sensonics, Inc., USA) ( Doty et al ., 1984 ), and did not report high levels of pre‐menstrual tension ( Steiner et al ., 1980 ). Of the 114 recruited women who met these inclusion criteria, 84 agreed to comply with study procedures and enrolled in the study.
Because preconceived ideas or knowledge about pheromones could potentially influence their responses, study participants were blind to the hypotheses and the source of the compounds. The study was presented to subjects as an examination of odour perception during the menstrual cycle; the word ‘pheromone’ was avoided in all communications. Participants were given a list of possible odorants that they might receive which included infant or sweat odours or no odour at all. They were told that the odorant might be different for each cycle. They were also asked to avoid wearing all perfumes or scented products for the duration of the study. All procedures were approved by the Institutional Review Board at The University of Chicago, and informed consent was obtained before participation.
All recipient women were studied for one baseline cycle during which they were given two vials daily, both of which contained a control pad ( Stern and McClintock, 1998 ). During the two subsequent experimental cycles, half of the women continued to be exposed to the control pads (control group; n = 30), whereas the remaining women were exposed to pads worn next to the axillae and breasts of lactating women, each of which was contained in a separate vial (experimental group; n = 27). Each woman in the experimental group received pads from at least three lactating women donors during each experimental cycle. Investigators who were blind to the identity of the donors and treatment condition of the subjects coded the vials containing the pads and generated the study’s block sequence for allocating women to study groups. At the end of a woman’s baseline cycle, when risk of study discontinuation had decreased (see Figure 1 ), these investigators also randomly assigned recipient women to the study groups using an algorithm based on the date of a woman’s next menses. A different set of investigators enrolled participants, distributed the pads to the subjects, and assessed the study outcomes. The latter investigators did not know the identity of the donor, the type of pad (i.e. breastfeeding, carrier control) offered, the cycle status of the recipient, or the method of group assignment.
Using methods established in our laboratory ( Stern and McClintock, 1998 ), recipient women were instructed to wipe the pads under their noses on the skin above their upper lips at least four times throughout the course of their daily activities during the course of the 3 month study. Each woman returned to the laboratory twice a week to receive the next set of vials and to ensure consistency in application procedures. These twice‐weekly visits were conducted in a small tiled windowless room (8×10 ft) with five room air changes per hour. The investigators who interacted with the participants were blind to whether they belonged in the experimental or control group. Under the supervision of the investigator, each applied the pads directly under their nose on the skin above the upper lip and was asked to respond in the affirmative or negative to the written question ‘Do you smell anything?’
Participants, who were trained on the procedures prior to the start of the study, recorded morning basal body temperature, evening cervical mucus characteristics and sexual activity, motivation and desire (the results of which will be reported elsewhere) throughout the 3 months of study participation. During the week prior to their expected day of ovulation, they tested their urine every evening (17:00–19:00) for LH (Ovukit; Quidel Corporation, USA). This method has >95% accuracy ( Stern and McClintock, 1996 ) in detecting the pre‐ovulatory LH surge, a singular hormonal event that triggers ovulation and demarcates the follicular from the ovulatory phase of the cycle. In addition, subjects collected first morning urine on the 5th, 7th and 9th days following the LH surge, from which we assayed for progesterone glucuronide, an indicator of ovulation and functional corpus luteum. These data were used together with data on vaginal secretions, menses, and basal body temperature to define the phases of the menstrual cycle ( Baviera et al ., 1988 ). Hereafter, menses is defined as the first to the last day of blood in vaginal secretions, including brown spotting. The follicular phase comprises the days between the last day of menses and the LH surge onset day. The ovulatory phase is the day of the LH surge onset plus the two subsequent days (ovulation occurs 30 h after the onset of the LH surge in urine). The luteal phase is comprised of the days between the end of the ovulatory phase and the first day of the subsequent menses. Menses onset demarcates the end of one menstrual cycle and initiates the next.
To assess the effect of breastfeeding chemosignals on normal menstrual cycles, we excluded women from analyses with abnormally long or short cycles that were self‐reported during the screening phase ( n = 2 women), experienced prospectively during the study (1 woman excluded for a long baseline cycle), or identified during preliminary analyses of initial cycle length (baseline, n = 7 women; cycle 1, n = 12 women). A normal cycle length was defined as 24–34 days, inclusive, which was within 1 SD of the study sample’s mean baseline cycle length. This is a standard inclusion criteria because cycles outside 1 SD of the study sample’s mean cycle length are known to be atypical and not likely to be followed by a normal cycle (14% of cycles at 18–35 years of age; Treloar et al ., 1967 ; Harlow et al ., 2000 ). More specifically, atypical cycles may reflect luteal phase defects or anovulation, and menses is more likely to result from an alternate mechanism such as the endometrial breakthrough bleeding ( Yen et al ., 1999 ).
The primary outcome variables were lengths of the three menstrual cycles and follicular and luteal phases of each cycle. To determine whether the baseline cycle length significantly predicted response to breastfeeding compounds when compared with control pads, cycle length data from each woman was plotted with respect to time, prior to any summary statistics. Next, to quantify the effect of initial conditions, we conducted simple linear regressions between the lengths of the initial cycle and subsequent cycles (i.e. between baseline length and cycle 1 length as well as between cycle 1 and cycle 2). Multiple regression was used to assess the effects of breastfeeding compounds on cycle length. Treatment condition (breastfeeding compound versus carrier control), length of the initial cycle, and their interaction were independent variables, and length of the subsequent cycle was the dependent variable. Similar models were used to assess effects of breastfeeding compounds on lengths of the follicular and luteal phases. For all of these analyses, the initial cycle (baseline or cycle 1) was normal in length (see above methods and definitions). A subset of women had normal cycles during both baseline and cycle 1 ( n = 42 women); an analysis of this subset revealed that breastfeeding compounds had the same significant effect.
From the data collected during the twice‐weekly visits, we determined how often each woman reported that she detected an odour (percentage of ‘yes’ responses obtained during the eight observation visits during baseline and 16 observation visits during the two treatment cycles). Group differences were tested using repeated measures of analysis on frequency of odour detection with treatment condition (breastfeeding, control) as the between‐subjects factor and cycle (baseline, experimental cycles) as the within‐subjects factor. Individual differences in reporting the pads’ odours were established with repeated measures analysis of variance (16 observations per participant). Within the breastfeeding group, we used multiple regression to determine whether women’s frequency of odour detection throughout the experimental months (16 observations per participant) mediated the effects of breastfeeding compounds on the ovarian cycle, with initial cycle length and odour detection frequency as independent variables and subsequent cycle length as the dependent variable. Power constraints precluded inclusion of other independent variables (e.g. ‘treatment condition’ and its interaction with other variables) into a larger multiple regression model that encompassed the control as well as the experimental groups.
Figure 1 summarizes participant flow through the study. Of the 84 study enrollees, 27 either withdrew or were disqualified during the baseline cycle, yielding the 57 women (control, 30; experimental, 27) who were allocated into study groups; of these, an additional three were disqualified during the first experimental cycle, leaving 54 women (control, 27; experimental, 27; all received the intended treatment) who completed all three cycles of the study. Reasons for disqualification included busy schedules (13 women), a baseline cycle ≥35 days ( n = 1), hormone ingestion ( n = 2), chicken pox ( n = 1), smoking ( n = 2), high pre‐menstrual scale rating ( n = 1), BMI >30% from ideal BMI ( n = 1), and 10 women lost to follow‐up during the baseline cycle. Of the 54 women who completed data collection throughout the study, 47 women (control, n = 22; experimental, n = 25) were included into the primary baseline to cycle 1 analyses, as per protocol, because they had an initial cycle within 1 SD (between 24 and 34 days, inclusive) of the group mean (mean baseline length = 29.2 ± 0.6 days, median = 29 days). These women were racially diverse (10% African American, 62% Caucasian, 2% Hispanic and 25% other) and were distributed into similarly aged groups (control: 23.2 ± 1.1 years; experimental: 25.1 ± 1.2 years; t = –1.2, P = 0.25). To assess whether the same effect was present under sustained exposure to breastfeeding compounds, 42 (control, n = 23; experimental, n = 19) of the 54 women who completed data collection throughout the study had a cycle 1 length between 24 and 34 days and thereby were included per protocol into the cycle 1 to cycle 2 analyses.
When cycle length data from each woman were plotted with respect to time (Figure 2 ), the striking difference between the experimental and control groups was the anticipated increase in cycle length variance that the treatment condition initiated. Moreover, in both conditions, the initial cycle length predicted the direction of change in subsequent cycle length.
Baseline cycles were typical in length for all women (mean = 29.2 ± 0.6 days, median = 29 days) and similar between groups (control: mean = 27.8 ± 0.5 days; experimental: 29.0 ± 0.6 days; t = –1.5, P = 0.15). Women in the control group had subsequent cycles that converged on the population modal length, a finding that is consistent with previous studies ( Harlow and Zeger, 1991 ; Spencer, 2001 ). In other words, baseline cycle lengths that were longer than the population mode shortened, whereas those that were shorter, lengthened during the subsequent cycle (Figure 2 A). As would be expected when cycles regress to the population mode, a woman’s initial cycle length did not predict the length of her subsequent cycle (see Figure 3 B and D). This lack of correlation was evident when comparing the length of the baseline cycle with that of cycle 1 ( r = 0.12, P ≤ 0.58) as well as when comparing cycle 1 with cycle 2 ( r = 0.11, P ≤ 0.60).
This normal pattern of variation was disrupted in those women who were exposed to breastfeeding compounds (Figure 2 B). Women in the experimental group whose baseline cycles were longer than the population mode either remained long or lengthened during the subsequent cycle, whereas those whose cycles were short either remained short or shortened. Indeed, many of the cycles after initial exposure to breastfeeding compounds became unusually long or short, tripling the overall group variance in cycle length. A woman’s cycle length during baseline significantly predicted the length of her subsequent cycle (baseline and cycle 1: r = 0.45, P ≤ 0.03). The change in cycle length from baseline to cycle 1 in the experimental group was significantly different from that of the control group (group×cycle length interaction in a multiple regression [ B = 1.33 (standardized regression coefficient = 3.18), P ≤ 0.05; Figure 3 A and B]).
The effect of breastfeeding compounds was even more pronounced during the second month of exposure (Figure 3 C and D). Cycle 1 lengths predicted cycle 2 lengths ( r = 0.82, P ≤ 0.001). Again, change in cycle length from cycle 1 to cycle 2 in the experimental group was significantly different from than that of the control group (group×cycle length interaction in a multiple regression [ B = 1.02 (standardized regression coefficient = 3.94), P ≤ 0.01]; Figure 3 C and D).
During the follicular phase, the experimental group showed the same response in phase length as seen for the overall cycle length. Baseline follicular phase lengths predicted the length of the subsequent follicular phase ( r = 0.70, P ≤ 0.001), and the follicular phase length in cycle 1 predicted its lengths in cycle 2 ( r = 0.86, P ≤ 0.001). Again, the pattern was significantly different from that of the control group during the baseline and first month of chemosignal exposure (group×cycle length interaction in a multiple regression [ B = 0.73 (standardized regression coefficient = 0.82), P ≤ 0.04]; Figure 4 A and B), an effect that became even more striking during a subsequent month of exposure (group×cycle length interaction in a multiple regression [ B = 0.94 (standardized regression coefficient = 1.0), P ≤ 0.003; Figure 4 C and D]. Therefore, the breastfeeding compounds also maintained follicular phase lengths according to an individual’s set‐point at the time of exposure and disrupted the normal pattern of regression to the modal length.
In the luteal phase, women exposed to breastfeeding compounds showed the same pattern of changes seen for overall cycle lengths. Baseline luteal phase lengths predicted the length of the subsequent luteal phase ( r = 0.39, P ≤ 0.05), and the luteal phase length in the first experimental month predicted its length in the second month of exposure ( r = 0.75, P ≤ 0.001). During the baseline and initial month of exposure, this pattern was significantly different from the control carrier (group×cycle length interaction in a multiple regression [ B = 1.24 (stnd coefficient = 1.43), P ≤ 0.04]; Figure 5 A and B) and there was a tendency for this difference to continue during the subsequent month of exposure (group×cycle length interaction in a multiple regression [ B = 0.64 (stnd coefficient = 1.22), P ≤ 0.07; Figure 5 C and D]. Therefore, both follicular and luteal phases drove the maintenance of individual cycles during the 2 months of exposure to breastfeeding compounds.
Women, on average, reported smelling an odour on approximately half the pads in each group (experimental group: 61.8 ± 6.4% SEM, control group: 46.8 ± 8.4% SEM). Women who were exposed to breastfeeding compounds were no more likely to report smelling an odour than women in the control group [ F (1, 45) = 0.46, P = 0.50]. Nonetheless, there were considerable individual differences within each group in how often women reported the presence of odour [experimental group: range: 10–100% of sessions, F (21, 334) = 11.057, P < 0.0001; control group: range: 0–100% of sessions, F (24, 373) = 6.198, P < 0.0001]. Frequency of odour detection, however, did not mediate the effect of the breastfeeding compounds on menstrual cycle length; when frequency of odour detection was added to the simple regression model for the experimental group, the previously observed relationship between initial and subsequent cycle length remained significant during exposure to breastfeeding compounds [cycle 1: baseline coefficient = 1.24 (stnd coefficient = 0.50), P < 0.01; cycle 2: cycle 1 coefficient = 1.05 (stnd coefficient = 0.74), P < 0.001]; moreover, there was no significant relationship of subsequent cycles (cycles 1 or 2) to frequency of odour detection [cycle 1: odour coefficient = –5.915 (standardized regression coefficient = 4.5), P = 0.21; cycle 2: odour coefficient = 2.68 (standardized regression coefficient = 2.7), P = 0.23]. Future studies should test smell detection more rigorously, although these results indicate that the detection of a smell did not mediate the effects of breastfeeding compounds on length menstrual cycle.
Exposure to breastfeeding compounds significantly disrupted the mechanism regulating the menstrual cycles and increased variance among women around the average cycle length. Consistent with animal model studies ( McClintock, 1984 ; Mennella and Moltz, 1989 ), such exposure tripled the variance of cycle length within the group. As anticipated, the effects of breastfeeding compounds depended on the type of initial cycles the women had (see Figures 2 and 3 ). Specifically, breastfeeding compounds maintained the type of cycle length an individual had at the time of exposure. That is, women who were having a long cycle continued to have long cycles; in addition, women who were having a short cycle continued to have short cycles, an effect not detected in rats because only an average change in cycle length was measured. Thus, breastfeeding compounds magnified individual differences in cycle length.
This effect became even more striking during a second cycle of exposure. In contrast, women in the control group experienced a variation in cycle length that was regulated towards the population modal cycle length, a finding that is consistent with previous studies ( Harlow and Zeger, 1991 ; Spencer, 2001 ). This dependence on initial conditions is one of the defining characteristics of a formally chaotic system ( Glass and Mackey, 1988 ) and further supports the contention that individual or subpopulation differences cannot be ignored when examining treatments that alter oscillating cycles such as the human menstrual cycle.
The observation that normal biological cycles have set points for their periodicity and that these set points can be altered in normal biological cycles is not new in the literature. The concept of ‘homeostasis of periodicity’ describes non‐random, oscillating systems ( Winfree, 1980 ) such as the human menstrual cycle. These systems run according to an endocrine ‘clock’ that remains in a particular state or returns to that state after having been in other accessible states resulting from perturbations ( Winfree, 1980 ). Inflicting a stimulus, such as a dose of a given hormone, or a chemosignal that triggers a hormonal response, should theoretically be able to reset the clock to a new set‐point, or disrupt the homeostatic regulation of periodicity, as we demonstrate here.
Because both the follicular and luteal phases of the cycle mediated the response to breastfeeding compounds, the neuroendocrine response to breastfeeding compounds could be mediated by LH pulsatility, estrogen threshold for the LH surge, as well as prolactin, progesterone and corticosterone secretion. Since the follicular phase responses and not luteal phase responses were significantly maintained during the second month of exposure, the present data suggest that more prepotent neuroendocrine mechanisms may be those predominant in the follicular phase. Animal model studies demonstrate that pheromones mediate ovarian cycle length by regulating the incidence of partial surges of LH as well as progesterone and prolactin levels after ovulation during the luteal phase ( McClintock, 1983a ,b; Gans, 1993 ). Nevertheless, this remains to be investigated further in humans, since inherent species differences in ovarian function preclude specific predictions about the human menstrual cycle based on the animal literature.
Although we found no evidence that odours emanating from the mother, infant, or both contributed to the observed effects on cycle length, research has shown that compounds emanating from a breastfeeding environment can be recognized by and affect the behaviours of both the mother and child ( Macfarlane, 1975 ; Cernoch and Porter, 1985 ; Schaal, 1986 ; Sullivan and Toubas, 1998 ). The newborns’ preference for their mothers’ breasts during conditions when they are unwashed and thereby more odorous ( Varendi et al ., 1994 ) suggest that, like other mammalian young, the recognition of and preference for maternal odours may play a role in guiding the infant to the nipple area and facilitating early nipple attachment and breastfeeding.
The present findings suggest that breastfeeding compounds have the potential for being social chemosignals and may or may not contain chemicals that ultimately fulfil the criteria for a human pheromone ( Beauchamp et al ., 1976 , 2000 ; McClintock, 2000 , 2002 ). The traditional definition of pheromones states that they are chemical signals produced by one member of a species and received by another, triggering neuroendocrine responses underlying behaviour and physiology ( Karlson and Butenandt, 1959 ; Karlson and Luscher, 1959 ). The active component(s) of the breastfeeding compounds modulated the neuroendocrine mechanisms controlling the timing of the pre‐ovulatory LH surge as well as the length of luteal phases of other women. These breastfeeding compounds are naturally produced, and contain axillary sweat and breast milk from the mothers, and most likely saliva from infants, and skin cells from either mother or infant.
That social interactions within groups of women and with men change the timing of the menstrual cycle and the pre‐ovulatory LH surge has also been demonstrated in a variety of settings. Other studies have also demonstrated that human axillary compounds from men and women regulate LH, its pulsatility, the timing of the pre‐ovulatory LH surge and menstrual cycle length ( Stern and McClintock, 1998 ; McClintock, 2000 ; Shinohara et al ., 2001 ; Preti et al ., 2003 ).
This direct application of compounds parallels the effect of spontaneous social interactions on the menstrual cycle. For example, women who spend time with men are more likely to have ovulatory menstrual cycles ( McClintock, 1971 ; Veith et al ., 1983 ; Stanislaw and Rice, 1987 ). In addition, women who live together have menstrual cycles that are either synchronized or non‐randomly distributed ( McClintock, 1971 ; A.Weller and L.Weller, 1993 , 1997 ; McClintock, 1998 ). It is important to acknowledge that not all investigations have found social interactions to affect ovarian cycles, although these latter studies focus on social settings where women live primarily with men, have infrequent interaction with each other, or have variable living conditions ( Graham, 1992 ; L.Weller and A.Weller, 1993 ).
These human effects parallel those seen in a wide variety of mammals. In other mammalian species, social chemosignals and pheromones also regulate ovulation ( McClintock, 2002 ). Among females with scarce resources, these systems can function to suppress ovulation in a form of competition or to signal the availability of resources. In humans, these compounds may have served to regulate fecundity in women in the context of traditional societies where limitations of food resources make fertility highly seasonal ( Ellison et al ., 2001 ). Women with variable menstrual cycle lengths (i.e. cycles that change by >10 days from cycle to cycle) have reduced fecundity (defined as the probability within a given cycle that a woman will become pregnant; Kolstad et al ., 1999 ). Prior to further speculation on function, more research needs to be done on the nature of the response when women are in a variety of natural reproductive states.
This is the first study to demonstrate that continuous exposure to breastfeeding compounds affects the neuroendocrine mechanisms regulating menstrual cycle length. Future studies should identify the specific neuroendocrine responses that mediate their marked effects as well as demonstrate these effects occurring in a normal social context—that is, with nulliparous women directly interacting with breastfeeding women as opposed to interacting artificially through collected, bottled secretions wiped underneath the nose. Future studies should also examine the effect of breastfeeding compounds on non‐normal cycles in which menstrual bleeding might be regulated by alternate mechanisms. Definitively labeling breastfeeding chemosignals as human pheromones will require demonstrating that they do indeed operate in the context of normal daily interactions with breastfeeding women and their infants.
This work was supported by a MERIT Award from the National Institute of Mental Health to MKM, R03 Dissertation grant from NIH to N.S., Monell Chemical Senses Center instituional funds to J.M.; S.J. was supported by NIH MD/PhD Training Grant and N.S. was supported by NIH Medical Scientist Training Program.
Figure 1. Flow diagram of study assessing effects of initial and continuous exposure of breastfeeding compounds on the menstrual cycle (experimental group versus control group). The diagram includes details of when enrolled women were excluded. The analyses were done only after all data were collected. In the main analysis, 22 women in the control group and 25 women in the experimental group fitted criteria for normal baseline cycles.
Figure 2. Change in cycle length between baseline cycle and cycle 1 for women in the control group ( A ; n = 22 cycles from 22 women receiving carrier control pads) and in the experimental group ( B ; n = 25 cycles from 25 women receiving breastfeeding pads). Symbols and lines represent individual women with normal baseline cycle lengths (24–34 days). Symbols and lines: upward‐pointing triangles and dashed lines represent women with a long baseline cycle length (>29 days), squares and smooth lines represent women with an average baseline cycle length (28–29 days), and downward pointing triangles and dotted lines represent women with a short baseline cycle length (<28 days).
Figure 3. Prediction of a woman’s menstrual cycle length by the length of her preceding cycle with linear regression. ( A ) and ( B ) are baseline to cycle 1. The effect of breastfeeding compounds was tested with multiple regression (bracket above A versus B ). ( C ) and D ) are cycle 1 to cycle 2 for women with a normal cycle 1; the effect of breastfeeding compounds was tested with multiple regression (bracket above C versus D ). The black circles represent individual women in the experimental group and the open circles represent women in the control group.
Figure 4. Prediction of a woman’s follicular phase length by the length of her preceding follicular phase with linear regression. ( A ) and ( B ) are follicular phases in baseline and cycle 1. The effect of breastfeeding compounds was tested with multiple regression (bracket above A versus B ). ( C ) and ( D ) are follicular phases in cycle 1 and cycle 2 for women with a normal cycle 1; the effect of breastfeeding compounds was tested with multiple regression (bracket above C versus D ). The black circles represent individual women in the experimental group and the open circles represent women in the control group.
Figure 5. Prediction of a woman’s luteal phase length by the length of her preceding luteal phase with linear regression. ( A ) and ( B ) are luteal phases in baseline and cycle 1. The effect of breastfeeding compounds was tested with multiple regression (bracket above A versus B ). ( C ) and ( D ) are luteal phases in cycle 1 and cycle 2 for women with a normal cycle 1; the effect of breastfeeding compounds was tested with multiple regression (bracket above C versus D ). The black circles represent individual women in the experimental group and the open circles represent women in the control group.
Source:
© 2022 - 2024
All rights reserved by
singlepheromones.com