The Effects of the Menstrual Cycle on Anterior Knee Laxity
- First Online:
- Cite this article as:
- Zazulak, B.T., Paterno, M., Myer, G.D. et al. Sports Med (2006) 36: 847. doi:10.2165/00007256-200636100-00004
- 272 Views
Female athletes are at a 4- to 6-fold increased risk of anterior cruciate ligament (ACL) injury compared with male athletes. There are several medical, emotional and financial burdens associated with these injuries. Sex hormones may be involved in the ACL injury disparity, with potential associations reported between phases of the menstrual cycle and ACL injury rates. The reported relationships between ACL injury and menstrual status may be related to associated changes in ligament mechanical properties from cyclic fluctuations of female sex hormones. A PubMed electronic database literature search, including MEDLINE (1966‐2005) and CINAHL (1982‐2005), with the search terms ‘menstrual cycle’ and ‘knee laxity’ was used for this systematic review. Studies were included in this systematic review if they were prospective cohort studies and investigated the association between the menstrual cycle and anterior knee laxity in females.
Nine prospective cohort studies, published as 11 articles, were included in the systematic review. Six of nine studies reported no significant effect of the menstrual cycle on anterior knee laxity in women. Three studies observed significant associations between the menstrual cycle and anterior knee laxity. These studies all reported the finding that laxity increased during the ovulatory or post-ovulatory phases of the cycle. A meta-analysis, which included data from all nine reviewed studies, corroborated this significant effect of cycle phase on knee laxity (F-value = 56.59, p = 0.0001). In the analyses, the knee laxity data measured at 10‐14 days was > 15‐28 days which was >‐9 days.
Future studies testing the relationship between the menstrual cycle and potentially associated parameters should consider the limitations outlined in this article and control for potential biases and confounders. Power analyses should be utilised. Subjects should be randomly entered into the studies at alternate points in the cycle, and standard and consistent data acquisition and reporting methods should be utilised. Future studies should clearly define what constitutes a ‘normal’ cycle and appropriate control subjects should be utilised. Furthermore, there is a need to define cycle phase (and timing within cycle phase) with actual hormone levels rather than a day of the cycle. Although hormone confirmations were provided in many of the studies that selected specific days to depict a particular cycle for all women, it is unknown from these data if they truly captured times of peak hormone values in all women.
A combined systematic review and meta-analysis of the literature indicate that the menstrual cycle may have an effect on anterior-posterior laxity of the knee; however, further investigation is needed to confirm or reject this hypothesis.
The cyclic changes in the circadian blood serum sex hormone levels are unique to the female endocrine system. The related sex hormones include estrogen, progesterone, relaxin and testosterone. The release of these hormones is orchestrated through a complex interaction among the hypothalamus, pituitary gland, ovaries and uterus. During the follicular or menstrual phase (days 1–9 of the menstrual cycle), estrogen, in a normally cycling woman, is secreted at a rate of approximately 60 µg/day. By the ovulatory phase (days 10–14 of the cycle), estrogen reaches a peak secretion rate of 400–900 µg/day, and decreases to approximately 300 µg/day during the luteal phase (day 15 to end of cycle). It is during the luteal phase that progesterone reaches its peak secretion rate of 25 mg/day making it the ovarian hormone with the highest secretion rate. These characterisations of the cycle are average levels of hormones that often demonstrate high variability among women. Relaxin levels rise in the follicular and luteal phases, peaking approximately 6–9 days after the luteinising hormone surge. Testosterone levels also fluctuate across menstrual cycle phases and contribute to the circulating concentration of estradiol by conversion via the process of aromatisation.
Sex hormones may affect the mechanical properties of the ACL, specifically collagen structure and metabolism.[15–21] Estrogen, progesterone, relaxin and testosterone receptors are present in human ACL tissue.[16,22–24] There is evidence to suggest effects of these hormones on the tensile properties of ligaments. Effects of ACL exposure to increased estrogen (specifically estradiol) levels include decreased fibroblast proliferation and reduced procollagen synthesis in cell cultures,[16,20] and a reduced load to failure rate in animal models.[15,18] Conversely, ACL exposure to increased progesterone is associated with increased fibroblast proliferation and collagen formation in cell cultures. Although estradiol and progesterone are the primary focus of previous research on hormonal influences on ligament properties, fluctuating levels of other hormones may also play a role in ligament behaviour. These potentially interrelated hormones include relaxin, which decreases soft tissue tension, endogenous testosterone,[23,25] estrone and estrial (two other circulating estrogens) and sex-hormone-binding globulin.
The potential influence of these sex hormones on the physical properties, specifically tensile strength and laxity, of the ACL in women is not delineated. However, it has been demonstrated that women generally have greater knee laxity than men.[27–29] Knee laxity (measured by knee hyperextension and generalised joint laxity) has been identified as a risk factor for ACL injury.[30–32] Several studies have examined the possibility that acute transient changes in knee laxity across the menstrual cycle may be a function of changing sex hormone levels.[23,25,26,33–40] The purpose of this article is to systematically review the literature regarding the association between menstrual cycle and anterior tibiofemoral motion of the knee as an indicator of ACL laxity.
There is controversy regarding menstrual phase terminology, specifically how cycle phases should be designated. For example, we have utilised the term ‘ovulatory phase’ to match the terminology used by all nine studies included in the systematic review.[23,25,26,33–40] However, ovulation is a point in time, not necessarily a ‘;phase’ of the menstrual cycle. It may be more appropriate to simply break the cycle into two phases: pre-ovulatory and post-ovulatory. It is important to note that all nine studies reported their female subjects were either eumenorrheic or had a ‘normal’ cycle phase, which was defined.[23,25,26,33–40] The ‘normal’ phase ranged from 24 to 35 days and the methods used to determine cycle phase varied from subject recall, day of cycle with hormone confirmation,[25,26,35,37–39] or hormone data alone. There is also a discrepancy in the terminology used by Heitz et al. compared with the other studies. Heitz et al. referred to the day 1 measurement as ‘menstrual‘; phase, not follicular and they referred to their day 11–13 measurements as captured during the follicular, not ovulatory phase. For purposes of consistency, we compare their ‘menstrual‘; phase measure with the follicular phase of the other eight authors and we compared their ‘follicular’ phase measure with the ovulatory phase designated in the other studies (table I).
The nine studies (11 articles) retrieved varied in the population of subjects and the method of determining the phase of menstrual cycle/hormone levels. Subjects consisted of collegiate athletes,[33,34] high-school athletes, athletes participating in various levels of sports and recreational activities,[26,38] non-athletes[37,40] and unspecified sports participation status.[35,39] The cohort sizes ranged from 7 to 41, with a median of 18. The anterior tibiofemoral motion was measured in all of the included studies with an anterior-directed force on the tibia (with a KT1000™ or KT2000™ arthrometer)1 with the subjects supine and the knee positioned at 25° of knee flexion. In addition, the reported data had large standard deviations for most studies. This variability may have masked potential positive effects. Worth noting, the Karageanes et al. and Belanger et al. studies did not numerically report the standard deviations of their data. Only one study randomised testing order by phase, one used a balance design for testing in the follicular and luteal phases, four studies randomly entered subjects but did not randomise testing by phase,[34–36,39] and three did not randomise subjects at all (started testing all subjects in the same phase [table I]).[1,7,33,38] The nine studies, published in 11 articles, are reviewed in sections 2.1–2.9 by order of publication date.
2.1 Study 1: Heitz et al.
This study was a prospective cohort study designed to determine if female recreational athletes experienced significant differences in knee laxity concomitant with the estrogen and progesterone surges during a normal menstrual cycle. The small cohort of female patients (n = 7) had a mean age of 26.9 years, no history of taking oral contraception medication and had a self-reported ‘normal’ (28–30 days) menstrual cycle. There was no control group (oral contraceptive female, pre-pubertal female, menopausal or matched male comparative group) for this study.
Baseline levels of estrogen and progesterone, measured on day 1 (onset of menses) of the cycle, were utilised as a baseline measurement of hormonal concentrations (phase I). Peak estrogen levels occur between days 10 and 13, which was categorised as the follicular phase (phase II). Progesterone level peaks at days 20–23, which was labelled the luteal phase (phase III). A venous blood draw was administered to each subject to assess circulating levels of estrogen and progesterone on days 1, 10, 11, 12, 13, 21, 22, 23 and 24. Immediately following each blood draw, each subject was assessed for knee laxity with the KT2000™ knee arthrometer at 67N, 89N and 133N of force. Knee laxity at 133N of force was analysed and reported. The authors reported a significant increase (p = 0.048) in knee laxity between phase I and phase II (peak estrogen surge), and a significant increase (p = 0.006) in knee laxity between phase I and phase III (peak progesterone surge).
The most notable limitations of this study were the size of the study group (n = 7) and the lack of a matched control group. In addition, the authors did not stipulate whether an experienced examiner was utilised, random assignment or counterbalance of subjects prior to data collection were utilised. Every subject was tested at menses and then sequentially during the later stages. Wroble et al. reported that subjects who were tested over several trials on different days with the KT arthrometer had the least translation on the first trial and then increased and plateaued translation during subsequent trials. They suggested that the KT arthrometer can be uncomfortable, which may cause protective muscular guarding tension that resists anterior tibial translation. Wroble et al. attributed their findings to a learning or ‘comfort’ effect of being exposed to the KT arthrometer. Since the single initial value of KT laxity at the onset of menses was the lowest in the subjects tested by Heitz et al., it is possible that without a counterbalanced subject testing, the demonstrated effects between the first and the subsequent measures could have been due to the described ‘comfort effect’ rather than sex hormones or cycle phase.
The authors attempted to target multiple points in time around the onset of menses and near the projected peak of estrogen and progesterone. However, as Shultz et al. have shown, changes in laxity may occur several days after fluctuations in hormonal concentrations occur. The authors did attempt to account for this variability by sampling multiple days during the target points to offer a better opportunity to capture laxity changes at critical points in the cycle than would be accomplished on a single test day; however, the data collection points targeted the peaks in hormones as opposed to days following peaks. A further limitation is that the authors did not state how they defined the term ‘normal’ menses. Ultimately, these study limitations may have decreased the potential to elucidate a conclusive relationship between cycle phase and laxity.
2.2 Study 2: Karageanes et al.
This prospective, single-blind cohort study monitored 26 female high-school athletes (mean age 15.7 ± 1.0 years) during a 5- to 8-week period in order to collect data from each phase of one complete menstrual cycle. The stated objective of this study was to determine if a significant change in laxity of the ACL occurs in the competitive adolescent female athlete throughout different phases of the menstrual cycle. The cohort included high-school females who participated in gymnastics, soccer, track, tennis and basketball with no comparative control group. The population of females had normal menstrual cycles.
A KT1000™ arthrometer was used to measure anterior tibiofemoral laxity (89N) during repeated measures over an 8-week period. Each subject was tested prior to workouts or competition at discontinuous intervals. The mean number of measurements taken per knee was 12.8, with a range between 6 and 21. The range of time between measurements was 1–7 days. The athletes charted menstrual periods on a monthly calendar that was submitted after testing in order to minimise bias in laxity readings.
The three phases of the menstrual cycle were calculated from the questionnaires. The authors counted 14 days prior to the first day of menses to estimate ovulation, and counted 3 days back to represent the ovulatory phase. The days from the beginning of menses to the beginning of ovulation represented the follicular phase, and the time from the estimated ovulation day to the first day of menses was designated as the luteal phase. The mean laxity measures for each phase were statistically compared (table I). The authors reported no significant difference in knee laxity throughout the three phases of the menstrual cycle (p > 0.05).
One limitation of this study was using a questionnaire to estimate the estrogen surge that occurs during ovulation. This is more accurately done through daily serum assay to account for the individual variability in fluctuating sex hormone levels. Moreover, the authors relied on self-report measures to describe the previous menstrual cycles, even though the correlation between blood hormone levels, cycle phase and self-report of phase is poor. The authors’ definition of ‘normal’ was the subject’s report of a cycle of 26–30 days with menses 4–7 days over the past 6 months. This may not be an adequate method for high data reliability and reproducibility especially given the inherent cycle variability in adolescent populations such as those tested. Furthermore, the population in this study was notably younger than the subjects included in the other studies.
It is not clear which laxity measurements were selected and extracted for the final data for analysis. In the measures that were selected, there was high variability in the number of measurements taken (range 6–21 measurements), as well as the range of time between measurements (1–7 days). The variability of the data measures was not reported, therefore it was difficult to interpret the relative coefficient of variation and whether there was potential beta error in the findings. The data were sampled randomly within each phase, which would make the presumption that each day within the phase is representative of a particular hormone milieu. Another potential limitation is that two examiners made the laxity measures and only measured at 89N (seven of the remaining eight studies measured laxity at 133–134N of force). There tends to be relatively high inter-rater variability in KT arthrometer measurements.
2.3 Study 3: Deie et al.
This prospective cohort study evaluated anterior-posterior tibiofemoral laxity changes in women during their menstrual cycle. The authors studied a cohort of 16 young women (mean age 21.6 years) regular menstrual cycles (28 ± 4 days) and no history of oral contraceptive use, and a control group of eight young men (mean age 21.5 years). Data collection in the study group occurred over a 4-week span. The data collected included a self-assessment of daily basal body temperature, weekly assessments of estradiol and progesterone levels via blood serum and two to three assessments of knee laxity per week using a KT2000™ arthrometer, administered by a single tester.
The control group was assessed three times per week with the KT2000™ for three consecutive weeks. No assessments of hormonal levels were assessed in the control group. For the study group, each knee laxity assessment was grouped into the follicular phase, the ovulatory phase or the luteal phase, based on the basal body temperature and the levels of estradiol and progesterone in the subject’s blood. The control group’s data were grouped by week. The authors reported significant differences in knee laxity between the follicular phase and the luteal phase at 134N and differences between the follicular and both the ovulatory and luteal phase at 89N (p < 0.05). The authors reported no significant difference in anterior knee laxity between the test periods for the control group.
The description of the methods used in this study was not sufficient to make accurate comparisons to other cited methodological approaches for phase categorisation of knee laxity data. The authors did state that each subject was measured for basal body temperature and levels of estradiol and progesterone, but there is no description of how these measures were used to classify each female subject into an appropriate menstrual stage. Secondly, the interpretation of the findings is limited, as the authors do not report the timepoint at which each participant began testing (such as at the onset of menses). Another limitation was the timing between the relationship of knee laxity testing and the estradiol and progesterone assessment. The authors reported that knee laxity was tested two to three times per week, but no routine cycle of testing was reported. Similarly, the concentrations of estradiol and progesterone were only assessed weekly. Considering that fluctuations in hormone concentrations are variable, and can dramatically change in 2–4 days, their weekly assessment may not have accurately classified the stages of the menstrual cycle for each subject. The testing was not randomised nor counterbalanced by phase and the knee laxity testing in the control group did not temporally match the testing frame of the female subjects. Lastly, the subjects’ sports status was not defined, which may limit the generalisability to the high-risk sports population.
2.4 Study 4: Arnold et al.
This study represents a prospective cohort study examining the relationship between serum relaxin levels and joint laxity in female athletes. The cohort of college-aged female athletes (n = 57) was subdivided into 41 uninjured athletes, eight non-athletes and eight ACL-injured athletes with a mean age of 19.3 + 1.5 years. A control group of five males was included. The authors may not have adequately controlled for use of oral contraceptives, menstrual history or the onset of menses in their methods. Relaxin levels were measured weekly for 4 weeks through venous blood assessment. Knee laxity was assessed at the time of the weekly blood draw, via a KT1000™ knee arthometer, administered by a single examiner. Anterior tibiofemoral translation was assessed at 67N (15lb), 89N (20lb) and manual maximum force; however, only maximum force was utilised for data analysis.
The authors reported no sex differences in mean levels of relaxin, but noted significant fluctuations in relaxin levels in women, week to week. They reported a trend towards increased knee laxity in women compared to men (especially in the injured female athlete group), no change in knee laxity measures through the course of a menstrual cycle and no significant correlation between laxity and relaxin levels (table I). The authors concluded that there was not a significant relationship between relaxin level and knee laxity.
Determining a relationship between a specific point in the menstrual cycle and knee laxity was difficult because of the lack of association of the days of testing to any stage of the menstrual cycle. Several authors have reported changes in hormonal levels during specific stages of the menstrual cycle. However, this study failed to link testing points to a significant cyclical event (i.e. onset of menses). Secondly, the testing was only executed weekly. Data from other authors suggest changes in knee laxity several days after changes in hormonal levels. Testing only once per week as opposed to daily is a concern as this would not be frequent enough to detect subtle changes in knee laxity after onset of menses. As previously noted, the testing of 1 day may be problematic as it is difficult to discern whether this 1 day in each phase is testing the same hormone environment.
The authors attempted to control for the use of oral contraceptives. However, as they noted, only 82% of the participants completed a questionnaire that determined the use of oral contraceptive use among the female participants. This is a potentially confounding variable, especially given that there is an uncertainty regarding the use of oral contraceptives in the remaining 18% and thus, the classification of these subjects. Interpretation of the hormonal effects on the ACL is confounded with the use of maximum force KT testing as the secondary restraints may contribute increased resistance at increased amounts of anterior force. Finally, the variability of the data was relatively high, which may have lead to potential beta error.
2.5 Study 5: Van Lunen et al.
This controlled laboratory cohort study monitored 12 mildly to moderately active females during three points of one menstrual cycle. The stated objective of this study was to determine whether ACL laxity was associated with concentrations of reproductive hormones during the menstrual cycle. The cohort of 12 ‘mildly to moderately active’ females was included in the study with no control group. The population of 12 females (mean age 24.3 ± 4.9 years) had ‘normal’ menstrual cycles of 28–35 days over the 12 months prior to the study. Subjects were tested at onset of menses, near ovulation and day 23 (mid-luteal phase). At each session, blood was drawn for radioimmunoassay and anterior tibiofemoral laxity was measured with KT2000™ (133N). One-day measures within each phase of the menstrual cycle were examined and statistically compared (table I). The authors reported no associations between follicular, ovulatory and luteal phase hormonal concentrations and anterior tibiofemoral laxity (p < 0.05).
This study attempted to test laxity and sex hormone levels near significant landmarks throughout the menstrual cycle. As the authors acknowledged, there was a variable time delay between initiation of menses (between 16 and 35.5 hours) or ovulation (9.75 and 35 hours) and the laxity measurements or blood draws that may have influenced the accurate timing between cycle phase, hormonal levels and these measurements. They attempted to account for this time delay by characterising mid-cycle measurements as ‘near ovulation’. One examiner made most, but not all, of the laxity measures (11 of 12), which permits increased potential inter-rater error. However, the reported data demonstrated low variability.
All of the subjects in this study began testing at the reported onset of menses. However, unlike Heitz et al., their consistent order of testing did not result in significant differences in knee laxity between cycle stages. This may be due to the formalised training session that the examiner underwent prior to testing. Consistency of menstrual cycle length prior to the study was self-reported rather than documented by the investigators. However, this issue may be mitigated by the fact that ovulation kits and actual hormone assays were used to determine ovulation. Lastly, the current description of the population ‘mildly to moderately active’ females may limit the comparison with female athletes who are at increased risk of ACL injury.
2.6 Study 6: Belanger et al.
This controlled laboratory cohort study monitored 18 female high-level collegiate athletes (age not defined) 2 times/week for 10 weeks. The stated objective of this study was to determine whether anterior tibiofemoral laxity was associated with concentrations of reproductive hormones during the menstrual cycle. The authors hypothesised that anterior tibiofemoral laxity would be significantly different in the follicular, ovulatory and luteal phases of the menstrual cycle. Knee laxity was measured by a single examiner using a KT2000™ (134N). Menstrual cycle phases were determined by charts of waking temperature and menstruation.
The initial cohort of 27 females had normal menstrual cycles and no history of amenorrhoea. However, the authors did not provide a definition of ‘normal’. Data from seven subjects were dropped due to inadequate compliance, whereas two subjects were dropped due to a failure to menstruate over the 10-week testing period. Individual cycle lengths were normalised to a 28-day cycle and divided into three phases: follicular, ovulatory and luteal. Knee laxity data were grouped according to these three phases and statistically compared (table I). The authors reported no significant differences in anterior tibiofemoral laxity in any of the three menstrual phases, before or after exercise (p < 0.05).
The authors acknowledged several limitations to this study. One was the high drop-out rate (data from nine of the initial 27 subject cohort were excluded, seven for non-compliance with the protocol and two because of oligomenorrhoea). Although these methods are commonly used to track ovulation, there is potential error and inconsistency in the methods of self-reported time of menstruation as well as the method of using waking temperature to detect ovulation. Another notable limitation of the study is normalisation of the menstrual cycle to 28 days via proportional scaling (although the authors did perform sensitivity analyses to determine if this normalisation affected the results). Furthermore, although the authors sampled multiple days, they did not attempt to sample around critical events. Even within each of the described phases, there are fluctuations in hormones from day to day. Thus treating all measurement days within a general phase (and not necessarily time around a particular event) as representative of the same may compromise the ability to capture peaks and valleys in the laxity data. The variability of the data measures was not reported, therefore it was difficult to interpret the relative coefficient of variation and whether there was potential beta error in the findings.
2.7 Study 7: Romani et al. and Lovering and Romani
This prospective cohort study was published in two separate reports.[23,26] The first identified the relationship between hormone levels and ACL stiffness. The second included free and total testosterone into the original statistical model. The cohort included 20 active, healthy female subjects (mean age = 25.9 ± 5.1 years) with menstrual cycles reported to be between 28 and 32 days long for the 3 months prior to the study. All subjects participated in an introductory session with the KT2000™ knee arthrometer test and were randomly assigned into three groups to begin data collection at the onset of menses, near ovulation or during the luteal phase of their menstrual cycle. At each stage of the menstrual cycle, three measurements of anterior tibial displacement were made with the KT2000™ and blood was drawn for assay analysis of sex hormone concentration during a single testing session. The onset of menses was defined as day 1 of the menstrual cycle. Data collection near ovulation was within 24 hours of positive testing with an ovulation kit and measurements during the luteal phase were taken between days 22 and 24. Hysteresis curves were used to determine stiffness between 89N and 134N. In order to provide a similar comparison between studies, knee laxity was calculated in a post hoc analysis of the same subjects and KT2000™ measurements.[23,26]
The means of knee laxity (table I) [menses: mean = 5.8 ± 1.6mm; near ovulation: 5.7 ± 1.8mm; luteal: 6.1 ± 1.7mm] did not significantly change between the three stages of the menstrual cycle. However, Spearman rank (rs) order analysis indicated a significant negative relationship between estradiol concentration and stiffness (rs = −0.70, p < 0.001) and a significant positive relationship between testosterone (rs = 0.48, p = 0.03) and free androgen index (rs = 0.44, p = 0.05) and stiffness. There were no significant relationships between any of the sex hormones and laxity. A Spearman partial rank (rsp) order analysis was used to determine the relationship between individual variables and knee laxity and stiffness while controlling for the influence of the other sex hormone variables. Estradiol was the only sex hormone that had a significant relationship with stiffness (rsp = −0.80, p < 0.001) indicating that estradiol was the only independent predictor of stiffness.
A limitation of this study was that the data were only collected over three consecutive menstrual stages during a single menstrual cycle. Thus, it is not known whether the relationships between sex hormone concentrations and measurements of ACL tensile strength over a longer period of time existed. As the authors pointed out, the term ‘ACL stiffness’ was used to describe the stiffness of the ACL and the other capsuloligamentous and musculotendinous structures that also play a role in restraining anterior tibial translation. In addition, the reported data had large standard deviations. This variability may have masked potentially significant differences in knee laxity between cycle phases.
2.8 Study 8: Shultz et al.
This prospective cohort study, published in two separate reports,[25,40] monitored non-athletic women during 20 different days of one menstrual cycle. The stated objective of this study was to quantify, through daily serial measures, changes in knee laxity as a function of changing sex hormone levels across one complete menstrual cycle. The cohort of 25 ‘non-athletic’ women (n = 22 in the laxity study) were included in the study while 20 men were used as a control group. The population of 25 women (mean age 23 ± 3.5 years) had normal menstrual cycles. The authors’ definition of ‘normal’ was the subjects’ report of a 28- to 32-day menstrual cycle over the past 6 months. Blood was drawn daily in women and once a week in men for serum assay of estradiol, progesterone and testosterone.
One-day measures within each phase were examined and statistically compared (table I). Data were not aligned by day of the cycle but rather by the actual changes occurring in hormone concentrations. Furthermore, they sampled 5 consecutive days in each phase. The authors found knee laxity was significantly greater on day 5 of the 5 days measured near ovulation when compared with day 3 of the 5 days measured at menses, and days 1–3 of the 5 days sampled in the early luteal phase compared with all 5 days of menses and day 1 of the 5 days measured near ovulation. The first article reported that the cyclic differences in knee laxity in this group of women correlated to concentrations of all three hormones, based on the cycle phase. Additionally, knee laxity changed 3, 4 and 5 days after changes in estradiol, progesterone and testosterone levels, respectively.
One limitation of the study was that the males had each measurement (laxity, blood draw with hormone of estrogen, progesterone and testosterone) taken on four test days, once per week, but a ‘single representative value for each variable across the four tests days’ was used for statistical comparison with the female data. In order to avoid the ‘comfort effect’ reported by Wroble et al. the authors used a counterbalanced subject test assignment to begin and end data collection at three stages of the menstrual cycle.
The authors used a self-report of previous menstrual cycle consistency for inclusion into the study, but measured sex hormones daily to determine cycle stage. In addition, the definition of ‘non-athletic’ is unclear. However, because the subjects were reported to be ‘non-athletic’ women, we may not be able to generalise these findings to other athletic female populations at high risk for ACL injury. The authors acknowledge the potential compromise in KT2000™ test reliability by using two examiners.
2.9 Study 9: Beynnon et al.
This controlled laboratory cohort study monitored 17 eumenorrheic women during 5 specific days of one menstrual cycle (early follicular, late follicular, mid-luteal, late luteal and repeat of early follicular) and were compared with 17 men. The stated objective of this study was to determine whether estradiol and progesterone levels are associated with increased anterior knee laxity. The population of 17 women (mean age 21.7 years, range 17–29 years) was described as eumennorrheic, demonstrating a normal monthly menstrual cycle. Anterior-posterior knee laxity (KT1000™) and serum concentrations of estradiol and progesterone were measured in the women at the five aforementioned timepoints matched with corresponding time intervals to the male controls.
During the menstrual cycle before testing, the women identified the first day of menses, used an ovulation test to document the day of ovulation and identified the day of the next cycle. For subjects with a 28-day cycle length, the testing was performed between day 1 and 3 (early follicular), between day 11 and 13 (late follicular), between day 20 and 22 (mid-luteal), between day 27 and 28 (late luteal) and a repeat of day 1–3. Laxity measures within each phase were examined (late follicular, cycle days 11–13, was designated as ovulatory phase for comparative purposes) and statistically compared (table I). The authors reported no significant difference in knee laxity across the menstrual cycle in women and no change over time in men (p > 0.05). There was no relationship between estradiol and progesterone fluctuation and knee laxity (p > 0.05). The authors reported greater knee laxity values in women compared with men (p = 0.01), consistent with previous studies.[27–29]
One limitation of the study is the high rate of exclusion (11 were excluded due to an anovulatory cycle), leaving 17 eumenorrheic females. Furthermore, the female data were not randomised (which, as described earlier, may have effects on serial KT testing), nor were the women randomly entered into testing. Another potential restriction for result interpretation was the self-report of menstrual history. In addition, the use of ‘non-athletic’ women may limit the ability to generalise these findings to the athletic female population, which is the population at increased risk for ACL injury. One examiner made most, but not all, of the laxity measures. There tends to be relatively high inter-rater variability in KT arthrometer measurements. Moreover, the authors did not document the menstrual cycle during the month of study with an ovulation kit. Considering variability between monthly menstrual cycles, this method does not ensure that the cycles are the same from one cycle to the next and that the measurements were performed at peak estradiol concentrations. Lastly, the inter- and intra-rater error may be increased with the KT1000™ knee arthrometer compared with more recent arthrometers.[44,45] This study did not report variability in the text; however, error bars in the figures were indicative of low variability and decreased chance of beta error.
Six of nine studies reported no significant effect of the menstrual cycle on ACL laxity as measured by knee anterior motion using an instrumented arthrometer.[26,33–37] However, the majority of studies that did not find an effect either based their findings on a single sampled day of the cycle, or randomly sampled across the cycle without hormonal or cycle landmark confirmation. This approach makes the assumption that within a cycle all days represent an equitable hormone milieu. Individual variation in hormonal status was likely the greatest confounding factor that obscured potential positive findings. Furthermore, some women’s ligaments may be more responsive to hormones than others, and this individual variation may also have masked possible significant findings. However, despite these potential sources of beta error, three studies did show significant associations between the menstrual cycle and anterior knee laxity.[25,38–40]
Three of the nine authors, Heitz et al., Deie et al. and Schultz et al.,[25,40] observed significant positive effects of the menstrual cycle on knee laxity (table I). Interestingly, these studies all reported the same basic findings; that laxity increased during the post-ovulatory phases of the cycle. This consistency in finding is compelling considering the inter-individual differences in hormonal fluctuation during a so-called ‘normal’ cycle. Although these three studies make similar conclusions, there are consistent limitations to these studies that limit the interpretation of their positive findings. Two of these three studies are the earliest and most weakly designed studies of those reviewed.[38,39]
A meta-analysis, which included data from all nine reviewed studies, demonstrated a significant effect of cycle phase on knee laxity (F-value = 56.59, p = 0.0001). The laxities measured at the three menstrual cycle times were significantly different. In the analyses, the knee laxity data measured at 10–14 days was >15–28 days, which was >1–9 days. However, the power to demonstrate these differences is greatly increased by pooling the study data and increasing the overall sample number. This analysis is a very rudimentary approach because the data represent repeat measures and we do not know the intercorrelations of the three knee laxity outcomes, since we grouped mean values for the data. It is difficult to determine whether we would arrive at the same conclusion if we were able to treat these data as repeated measurements in the meta-analysis.
All three studies that reported significant associations between the menstrual cycle and anterior knee laxity found the increased laxity during the ovulatory and post-ovulatory (luteal) phases. The observed increase in anterior translation does not coincide with the majority of the published studies regarding increased ACL injuries during the pre-ovulatory to ovulatory phases of the menstrual cycle (figure 1).[1,6–10] The timing of increased laxity during mid-cycle reported in these three studies[25,38,39] is consistent with the epidemiological studies by Wojtys et al.[9,10] who found an increased injury rate near mid-cycle. Shultz et al. discuss the hormonal influence on ligaments and speculate that there may be a possible delayed effect due to the turnover time of the collagen fibrils. However, a direct connection between anterior knee laxity and increased ACL injury risk is not well established in the literature.
Two additional studies, one that did find effects of the cycle on laxity and one that did not, examined the relationship between sex hormones and stiffness.[23,25,26,40] There was not a consistent association between stiffness and menstrual phase. However, there were significant associations between the hormone concentrations within an individual and that individual’s knee stiffness or laxity as calculated by an arthrometer measurement. In addition, despite being very well conceived studies, the calculation of stiffness used only two force values and was essentially a stiffness index, or the reverse of the compliance index often calculated with data from the KT2000™ to determine the integrity of the ACL. As such, the calculation of stiffness based only on this difference in force divided into the change in translation between the two forces is an important limitation of the studies.[23,26]
Individual variation in the hormonal milieu within a menstrual cycle and responsiveness to hormones may be a factor in the reviewed studies. It is possible that some women demonstrate large alterations in laxity through the cycle while others demonstrate little or none. This would explain the correlations observed between hormonal levels and knee stiffness, and lack of statistically significant group differences in anterior laxity observed across the specific phases of the menstrual cycle. Any individuality in how women respond to fluctuating concentrations of sex hormones would be especially important in female athletes with menstrual dysfunction or inconsistent cycles. If menstrual dysfunction occurs more frequently in female athletes than in the normal population of women, comparing the results of studies completed on non-athletic populations or on women with ‘normal’, consistent menstrual cycles may be unrealistic in a competitive athletic population. All nine studies in this review included only women with ‘normal’ cycles defined by some range of days for the cycle. This criterion resulted in a high number of subjects excluded due to irregular cycles or anovulatory cycles. The reported range of dysfunction is 3.4–66% in athletes, whereas rates of 2–5% have been reported in non-athletic women.
The nine studies reviewed varied widely in the athletic status of the women studied, which compromised the homogeneity of the results of our systematic review. However, several studies had to exclude a high number of women due to irregular cycles, which necessitates further investigations on athletic females to delineate potential menstrual cycle difference in athletes compared with those with cycles closer to the mean length. This would help determine if the results measured in ‘non-athletic’ populations may be generalised to female athletes who are at high risk of ACL injury. Ideally, future investigations would include women with varying cycle lengths compared with controls on oral contraception to provide a clearer picture of how sex hormones or menstrual cycle stage in a competitively athletic population may influence knee laxity and stiffness.
The timing of the potential effects of these sex hormones on the physical properties of the ACL is also not defined. For example, although collagen turnover is relatively rapid, it remains unknown how long it takes the ACL to remodel and to effect a change in strength, stiffness or laxity. Studies 7 and 8 reported relationships between estradiol and ligament mechanical properties. The high inter-individual variability in the levels of hormone concentration and ACL/knee connective tissue strength lead these authors to postulate that the relationship between sex hormones and strength/stiffness may be due to fluctuating concentrations of circulating hormones.[23,25,26,40] Therefore, a significant change in an athlete’s hormonal milieu (e.g. amenorrhoea, oral contraceptive use, menopause), may alter ACL strength characteristics after a period of exposure to that altered milieu. The rabbit data from Slauterbeck et al. support this theory. The treatment group of ovariectomised white rabbits was exposed to high levels of estrogen through silastic implants for approximately 1 month. Rabbits exposed to estradiol had a lower load to failure than controls without estradiol. Unfortunately, the timeframe required to alter the strength characteristics of the ligament and the potential relationship between ACL laxity and failure strength are to date uncharacterised.
One strength of this systematic review was that a KT1000™ or KT2000™ arthometer was used as the measurement device in all nine of these studies. However, a weakness of the comparison of these different data sets is that the KT1000™ uses no plotter and requires the examiner to compare tones (indicating a specific load) with values on an analogue dial. The KT2000™ uses the XY plotter with the latest derivation (the Compu-KT) using computer software. The literature demonstrates the KT is reliable if used by experienced examiners and with appropriate subject set-up and education. However, KT arthrometers have the potential to provide variable findings from the same individual subject. Wroble et al. demonstrated a modified ‘learning effect’, due presumably to muscular ‘guarding’ during initial testing. As the patient is repeatedly tested, they tend to relax and greater translation can result. Therefore, the finding of increases at later stages in the cycle may have been due to this effect, if the testing order among phases of the cycle was not randomised. Only one study randomised phase testing order, one counterbalanced, while seven of nine did not randomise testing order. Furthermore, the reliability of the testers was not consistently reported in all studies. Only one of the three studies[25,38,39] that found a statistically significant association between laxity and cycle phase reported reliability, while three[26,34,37] of six[26,33–37] that did not find an association reported moderate to good reliability.
There are several limitations to this systematic review. For example, there were nine different study designs that increased heterogeneity between studies. In addition, millimetres of laxity was not reported in all nine studies. Two of the studies chose not to report this parameter.[23,25,26,40] The two most recent studies published stiffness indices, rather than laxity measures.[23,25,26,40] The utility of a stiffness index versus laxity taken at a specific load (which is essentially stiffness at one force value) is that movement in the toe region may not be indicative of the tensile property but may be indicative of more about the length of the ligaments, capsule, etc. What is interesting is that when Shultz et al.,[25,40] eliminated the total laxity and used only the ‘change’ in laxity, they had a better fit for their model. This ‘change’ measure may actually be indicative of the tensile properties of later linear region of the length tension curve and may be more indicative of the property of the tissue than laxity or length alone. However, these authors are alone in their choice to look at stiffness.[23,25,26,40]
This systematic review of the literature indicates that the menstrual cycle may have a significant effect on anterior knee laxity. Although six of nine studies observed no significant effect of the cycle on ligament laxity,[26,33–37] this lack of positive evidence does not preclude potential effects of sex hormones on ligament integrity. A meta-analysis, which included data from all nine reviewed studies, demonstrated a significant effect of cycle phase on knee laxity (F-value = 56.59, p = 0.0001). The laxities measured at the three menstrual cycle times were significantly different, after taking into account the study and the force. In the analyses, the knee laxity data measured at 10–14 days was >15–28 days which was >1–9 days.
Interindividual variation in cycle hormone fluctuation may be the greatest challenge in performing unbiased studies with minimal confounding variables to develop consistent and valid conclusions. Most of the studies reviewed reported data that had relatively high standard deviations, while three studies did not report their inter-subject variability.[34–36] This variability may have masked potential positive effects. Interestingly, all three studies that observed differences in cycle phases reported increased laxity during the ovulatory and luteal phases relative to the follicular phase (table I).[8,13,38] This consistency of finding is intriguing and may indicate that there is a difference between the pre-ovulatory and post-ovulatory halves of the menstrual cycle. This finding shows a mixed relationship to the injury data (figure 1). The injury data are more indicative of injuries occurring during the pre-ovulatory half of the cycle,[19,48] when the ACL (or knee) would be less lax or demonstrate greater stiffness. However, the follicular phase data of increased knee laxity do not agree with the injury data. This contradictory relationship between the follicular phase laxity and injury findings may be further evidence that cycle-dependent changes in hormone concentrations may not consistently influence knee laxity, or possibly that more injuries occur when the ligament is stiffer rather than more lax. The effect may be variable between individuals. Alternatively, the effects of the menstrual cycle may be on the active restraints (neuromuscular in nature) rather than the passive restraints (ligament) on knee stability.
This systematic review of the literature shows evidence of a menstrual cycle effect on anterior-posterior laxity of the knee, though an unequivocal test of this hypothesis has yet to be performed. Future studies testing the relationship between the menstrual cycle and potentially associated parameters such as laxity, injury and neuromuscular control should consider the limitations outlined in this systematic review and control for the many potential biases and confounding factors. Power analyses should be utilised in order to ensure that the chance of beta error, or the acceptance of a negative finding that is actually positive, is minimised. In addition, measures that can have high inter-rater variability, such as knee arthrometer measures, should employ a single examiner. Appropriate control subjects should be utilised, for example, women on oral contraceptives, or pre-pubertal females who are not experiencing the cyclic effects of hormonal fluctuation, could be used as negative controls. The subjects should be entered into the studies at random times in the cycle and standard and consistent data acquisition and reporting methods should be utilised, including serum analysis of hormone concentrations for the determination of cycle phase to account for the wide cycle variability within and between individuals, which is especially relevant for the athletic population. Future studies should incorporate statistical models that do not require exclusion of individuals based on cycle length and randomly test for peak hormone concentrations. Using these methods, new studies may unequivocally determine the contributions of cyclic fluctuations of hormones to knee ligament laxity.
The use of trade names is for product identification purposes only and does not imply endorsement.
The authors would like to acknowledge funding support from National Institutes of Health Grant R01-AR049735-01A1 (TEH). The authors would like to acknowledge the assistance of Paul Succop, PhD, for statistical consultation with the meta-analysis, Thom Guidone, PT, Carrie-Lynn O’Donell, Tiffany Evans, Carmen Booth, DVM, PhD, and Jeanette Vitello, PT, for assistance with the preparation and review of the manuscript. The authors would also like to thank the authors of these nine studies for their contributions to answering this important question. The authors have no conflicts of interest that are directly relevant to the content of this review.