Efecto de la atorvastatina sobre los niveles de testosterona
Resumen
Antecedentes
Las estatinas son una de las clases de fármacos más prescrita a nivel mundial. La atorvastatina, la estatina más recetada, se utiliza actualmente para tratar afecciones como la hipercolesterolemia y la dislipidemia. Al reducir la concentración de colesterol, que es el precursor de la vía de la esteroidogénesis, la atorvastatina podría causar una reducción de los niveles de testosterona y otros andrógenos. La testosterona y otros andrógenos desempeñan papeles importantes en las funciones biológicas. Una posible reducción de los niveles de andrógenos, provocada por la atorvastatina, podría tener efectos negativos en la mayoría de los contextos. Por el contrario, en el marco del síndrome de ovario poliquístico (SOPQ), la reducción de los niveles excesivos de andrógenos con atorvastatina podría ser beneficiosa.
Objetivos
Objetivo principal
Cuantificar la magnitud del efecto de la atorvastatina sobre la testosterona total, tanto en hombres como en mujeres, en comparación con el placebo o ningún tratamiento.
Objetivos secundarios
Cuantificar la magnitud de los efectos de la atorvastatina en las concentraciones de testosterona libre, la globulina fijadora de hormonas sexuales (SHBG), la androstendiona, el sulfato de dehidroepiandrosterona (DHEAS), el índice de andrógenos libres (IAL) y los retiros debido a efectos adversos (RDEA), tanto en hombres como en mujeres, en comparación con el placebo o ningún tratamiento.
Métodos de búsqueda
El documentalista del Grupo Cochrane de Hipertensión (Cochrane Hypertension) buscó ensayos controlados aleatorizados (ECA) en las siguientes bases de datos hasta el 9 de noviembre de 2020: el Registro especializado del Grupo Cochrane de Hipertensión; el Registro Cochrane central de ensayos controlados (Cochrane Central Register of Controlled Trials; CENTRAL); MEDLINE; Embase; dos registros internacionales de ensayos y las páginas web de la Food and Drug Administration de EE.UU., la Oficina Europea de Patentes y la empresa farmacéutica Pfizer. En las búsquedas no hubo restricciones de idioma. También se estableció contacto con los autores de los artículos relevantes con respecto a otros trabajos publicados y no publicados.
Criterios de selección
ECA de atorvastatina diaria durante al menos tres semanas, comparada con placebo o ningún tratamiento, y que evalúen el cambio en los niveles de testosterona en hombres o mujeres.
Obtención y análisis de los datos
Dos autores de la revisión, de forma independiente, examinaron las referencias, extrajeron los datos y evaluaron el riesgo de sesgo de los estudios incluidos. Se utilizó la diferencia de medias (DM) con intervalos de confianza (IC) del 95% asociados para informar el tamaño del efecto de los desenlaces continuos y la razón de riesgos (RR) para informar los tamaños del efecto del único desenlace dicotómico (los RDEA). Se utilizó un modelo metanalítico de efectos fijos para combinar las estimaciones del efecto entre los estudios, y la razón de riesgos para informar el tamaño del efecto de los desenlaces dicotómicos. Para evaluar la certeza de la evidencia se utilizó GRADE.
Resultados principales
Se incluyeron seis ECA con 265 participantes que completaron el estudio y se informaron sus datos. Los participantes de dos de los estudios eran hombres con un perfil lipídico normal o con una dislipidemia leve (n = 140); la media de edad de los participantes fue 68 años. Los participantes de cuatro de los estudios eran mujeres con SOPQ (n = 125); la media de edad de las participantes fue 32 años. No se encontraron diferencias significativas en los niveles de testosterona de los hombres entre la atorvastatina y el placebo, DM ‐0,20 nmol/l (IC del 95%: ‐0,77 a 0,37). En las mujeres, la atorvastatina podría reducir la testosterona total en ‐0,27 nmol/l (IC del 95%: ‐0,50 a ‐0,04), el IAL en ‐2,59 nmol/l (IC del 95%: ‐3,62 a ‐1,57), la androstendiona en ‐1,37 nmol/l (IC del 95%: ‐2,26 a ‐0,49) y el DHEAS en ‐0,63 μmol/l (IC del 95%: ‐1,12 a ‐0,15). Además, en comparación con el placebo, la atorvastatina aumentó las concentraciones de SHBG en las mujeres en 3,11 nmol/l (IC del 95%: 0,23 a 5,99). No se identificaron estudios en mujeres sanas (es decir, mujeres con niveles normales de testosterona) ni en niños (menores de 18 años). Es importante destacar que ningún estudio informó acerca de la concentración de testosterona libre.
Conclusiones de los autores
No se encontraron diferencias significativas entre la atorvastatina y el placebo en la concentración de testosterona total en hombres. En mujeres con SOPQ, la atorvastatina redujo la testosterona total, el IAL, la androstendiona y el DHEAS. La certeza de la evidencia varió de baja a muy baja en ambas comparaciones. Se necesitan más ECA que estudien el efecto de la atorvastatina sobre la testosterona.
PICO
Resumen en términos sencillos
¿Cuál es el efecto de la atorvastatina sobre los niveles de testosterona y otras hormonas en hombres y mujeres?
Antecedentes
Las estatinas son una de las clases de medicamentos más prescritas y la atorvastatina es el medicamento más utilizado de esta clase. Las personas toman estatinas para reducir las concentraciones de colesterol en sangre y disminuir el riesgo de enfermedades cardíacas. Sin embargo, las estatinas también pueden reducir las concentraciones de testosterona y otros andrógenos. Se trata de hormonas que tienen funciones importantes en la salud y el desarrollo de hombres y mujeres.
¿Qué se hizo?
Se quiso medir el efecto de la atorvastatina sobre la testosterona y otros andrógenos.Se buscaron todos los estudios en la literatura médica que compararan el efecto de diferentes dosis de atorvastatina con placebo o ningún tratamiento en hombres y mujeres, y que también informaran sobre los niveles de testosterona y otros andrógenos. Se buscaron los estudios en los que los tratamientos que recibieron las personas se decidieron al azar. Este tipo de estudios suele proporcionar la evidencia más fiable sobre los efectos de un tratamiento. Los estudios podían incluir participantes de cualquier edad.
Tras identificar los estudios, se compararon los resultados y se resumió la evidencia de todos ellos. Finalmente, la confianza en la evidencia («certeza de la evidencia») se calificó sobre la base de factores como los métodos y los tamaños de los estudios, y la coherencia de los hallazgos entre los estudios.
¿Qué se encontró?
Se encontraron seis estudios con un total de 265 participantes. Los estudios se realizaron en China, EE.UU., Finlandia, Irán, Reino Unido y Turquía.
La evidencia de cuatro estudios indica que la atorvastatina podría tener un efecto potencialmente beneficioso en las mujeres con síndrome de ovario poliquístico (SOPQ), un conjunto de síntomas que pueden aparecer en mujeres con niveles de andrógenos más altos de lo normal. En mujeres con SOPQ, la atorvastatina ayudó a reducir la testosterona total y otros andrógenos.
Se encontraron dos estudios en hombres en los que la atorvastatina no tuvo un efecto significativo sobre la testosterona total.
¿Qué significa esto?
La certeza de la evidencia para todos los desenlaces varió de baja a muy baja. Las estimaciones estadísticas de los efectos de la atorvastatina sobre la testosterona y otros andrógenos en hombres y mujeres podrían ser muy diferentes de los verdaderos efectos. Se necesitan más estudios para responder a esta importante pregunta.
¿Cuál es el grado de actualización de esta evidencia?
La evidencia de esta revisión está actualizada hasta noviembre de 2020.
Authors' conclusions
Summary of findings
| Population: Male participants Intervention: Atorvastatin Comparison: Placebo or no treatment | |||
| Outcomes | Number of participants (RCTs) | Mean difference of atorvastatin compared to placebo (95% CI) | The certainty of evidence (GRADE) |
| Total testosterone | 140 (2) | ‐0.20 nmol/L (‐0.77 to 0.37) | Very low 1, 2, 3 |
| Acronyms and grades | |||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; RCT: randomized controlled trial | |||
| GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect | |||
| 1 Unclear risk of selection bias. 2 95% confidence interval around the best effect estimate includes both a reduction and an increase in total testosterone. 3 Evidence was derived from only two studies, which had several limitations. | |||
| Population: Adult female participants with PCOS Intervention: Atorvastatin Comparison: Placebo | |||
| Outcomes | Number of participants (RCTs) | Mean difference of atorvastatin compared to placebo (95% CI) | The certainty of evidence (GRADE) |
| Total testosterone | 125 (4) | ‐0.27 nmol/L (‐0.50 to ‐0.04) | Very low 1, 2,3 |
| FAI | 65 (2) | ‐2.59 (‐3.62 to ‐1.57) | Very low 1, 2, 3 |
| SHBG | 65 (2) | 3.11 nmol/L (0.23 to 5.99) | Low 1, 3 |
| Androstenedione | 125 (4) | ‐1.37 nmol/L (‐2.26 to ‐0.49) | Very low 1, 2, 3 |
| DHEAS | 125 (4) | ‐0.63 μmol/L (‐1.12 to ‐0.15) | Low 1, 3 |
| Acronyms and grades | |||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; DHEAS: dehydroepiandrosterone sulphate; FAI: free androgen index; RCT: randomized controlled trial; SHBG: sex hormone binding globin | |||
| GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect | |||
| 1 High risk of bias in source of funding and sponsorship bias in more than one study. 2 Significant heterogeneity. 3 Small sample size. | |||
Background
Androgens are a group of sex steroid hormones that are mainly responsible for the development and maintenance of the male sex organs, and secondary male sex traits. Androgens also impact libido and sexual behavior in both males and females. Circulating androgens include testosterone, dehydroepiandrosterone (DHEA), dehydroepiandrosterone sulfate (DHEAS), androstenedione (A4), androstenediol (A5), and dihydrotestosterone (DHT). Androgens are mainly produced by the testicles, adrenal cortex, and ovaries (Sriram 2010).
Testosterone, C19 H28 O2, is the main male sex hormone. It plays a crucial role in several biological processes, such as regulating the sex drive in both sexes and producing sperm in males. Testosterone is also involved in regulating some other functions in both males and females, e.g. bone mass, muscle mass and strength, fat distribution, and the production of red blood cells (Finkelstein 2013).
Testosterone circulates in the blood in three different forms: as testosterone firmly attached to sex hormone binding globulin (SHBG) (roughly 45% of total testosterone), which is biologically inactive; as free testosterone (approximately 2% to 3% of total testosterone); and as testosterone weakly bound to other proteins, mainly albumin (approximately 50% of total testosterone). Both free testosterone and testosterone attached to albumin are bioavailable for use by tissues (Dunn 1981).
Bioavailable and free testosterone are better measures than total serum testosterone to diagnose abnormal androgen status and their clinical sequelae. However, bioavailable and free testosterone tests are time‐consuming, expensive and not always available. Therefore, determining total testosterone level, which is easy and relatively inexpensive to do, is commonly used to diagnose overt male hypogonadism. However, it is an unreliable marker in patients who have total testosterone levels just below the normal range, or in the low normal range. To overcome this issue and provide a more reliable indicator, some researchers use the SHBG and total testosterone, with a formula to calculate an approximate estimate of bioavailable or free testosterone (Stanworth 2008).
The normal ranges of the serum total testosterone are significantly higher in adult men than in adult women, and vary widely depending on various factors that include the laboratory methods used and the patient's age. There is a significant increase in total testosterone in males after puberty and a reduction of approximately 1% in total testosterone per year in males over thirty years old (Brawer 2004; Feldman 2002; Pagana 2015; refer to Table 1).
| Tests of blood: normal findings | ||
|---|---|---|
| Free testosterone pg/mL | ||
| Age/Tanner stage | Male | Female |
| Postmenopausal | 0.6‐3.8 | |
| 7 months‐9 years (Tanner stage I) | ≤ 3.7 | < 2.2 |
| 10‐13 years (Tanner stage II) | 0.3‐21 | 0.4‐4.5 |
| 14‐15 years (Tanner stage III) | 1.0‐98 | 1.3‐7.5 |
| 16‐17 years (Tanner stage IV) | 35.0‐196 | 1.1‐15.5 |
| 18‐19 years (Tanner stage V) | 41.0‐239 | 0.8‐9.2 |
| Free testosterone % | ||
| Adult male | 1.6‐2.9 | |
| Adult female | 0.1‐0.3 | |
| Total testosterone ng/dL | ||
| Male | Female | |
| 7 months‐9 years (Tanner stage I) | < 30 | < 30 |
| 10‐13 years (Tanner stage II) | < 300 | < 40 |
| 14‐15 years (Tanner stage III) | 170‐540 | < 60 |
| 16‐19 years (Tanner stage IV, V) | 250‐910 | < 70 |
| 20 years and over | 280‐1080 | < 70 |
| Dihydrotestosterone | ||
| Adult male | 240‐650 pg/mL | |
| Adult female | ≤ 300 pg/mL | |
Total testosterone and free testosterone levels by sex and age (Pagana 2015).
Research suggests that low serum testosterone in men is associated with health problems, such as reduction in fertility and decreased libido. Low testosterone in both males and females can increase body fat and the incidence of depression. It can also decrease mass and strength of muscles, decrease body hair, decrease the production of red cells leading to anaemia, and decrease bone density (Demers 2010; Traish 2009).
Description of the condition
Statins are 3‐hydroxy‐3‐methylglutaryl–coenzyme A (HMG‐CoA) reductase inhibitors, a class of drugs mainly prescribed to lower blood cholesterol. They are widely used in adult patients for secondary and primary prevention of cardiovascular events (Stone 2014). They are also used to lower cholesterol in people of all ages with heterozygous familial hypercholesterolaemia and mixed dyslipidaemia (Schaiff 2008). They are considered to be a highly effective class of drugs in reducing low‐density lipoprotein (LDL) cholesterol (Schaiff 2008). Statins are also used to treat women with polycystic ovary syndrome (PCOS) (Sun 2015).
Description of the intervention
Atorvastatin is the most prescribed member of the statin class, which are among the most commonly prescribed medications worldwide (Ioannidis 2014; IQVIA 2017). In Canada, approximately 3 million people aged 20 to 79 (12% of Canada’s population) took a statin medication during the period 2007 to 2011; one in four Canadians are potentially eligible for statin treatment (Hennessy 2016). Studies illustrate that long‐term statin therapy in secondary prevention significantly reduces all‐cause mortality and major adverse cardiovascular events such as myocardial infarction (MI) and stroke. However, it remains controversial whether they reduce mortality in primary prevention patients (Virani 2013; Vrecer 2003).
Atorvastatin is rapidly absorbed after oral administration; reaching peak plasma concentration within one to two hours. Atorvastatin is extensively metabolized by cytochromes P‐450 3A4 and P‐450 3A5 to two active metabolites ortho‐ and para‐hydroxylated derivatives, which contribute to approximately 70% of the inhibitory activity for HMG‐CoA reductase. These two active metabolites also increase the half‐life of inhibitory activity for HMG‐CoA reductase (about 20 to 30 hours). A Cochrane Review shows that atorvastatin 2.5 mg/day to 80 mg/day reduces total blood LDL‐cholesterol by between 27.3% and 52%, respectively (Adams 2015; FDA 2011).
Statins are associated with a number of potential adverse effects. Such adverse effects can be divided into two groups, common adverse effects and serious adverse effects. Common adverse effects include headache, nausea, vomiting, constipation, diarrhoea, rash, insomnia, myalgia, dyspepsia, pain in extremities, upper respiratory infection, arthralgia, elevated creatine kinase (CK), elevated alanine aminotransferase (ALT) or aspartate aminotransferase (AST), cognitive impairment, and gynaecomastia. Serious adverse effects include muscle symptoms, ranging from muscle pain and weakness to rhabdomyolysis, tendon rupture, acute renal failure, hepatotoxicity, pancreatitis, hypersensitivity reaction, anaphylaxis, photosensitivity, toxic epidermal necrolysis, erythema multiforme, Stevens‐Johnson syndrome, thrombocytopaenia, leukopaenia, haemolytic anaemia, diabetes mellitus, and interstitial lung disease (atorvastatin Adverse Reactions ‐ Epocrates Online; Thompson 2016).
The most relevant adverse effect in terms of this review is gynaecomastia. This is listed as a common side effect of statins, which means it occurs in more than 1% of patients. It is a condition clinically defined by an enlargement of male breast tissue, along with other histopathological characteristics including benign proliferation of glandular male breast tissue (Johnson 2009). While the mechanism of action is unclear, statins could lead to gynaecomastia via an increased production of oestrogen, decreased production of androgens, or a combination of both. According to a recent case‐control study based on 6147 cases, the use of statins is associated with an increased risk of gynaecomastia (Skeldon 2018). The occurrence of gynaecomastia among male patients receiving statins most likely reflects an effect of statins to reduce androgen serum levels. This coincides with the hypothesis of our review.
How the intervention might work
The synthesis of steroid hormones occurs within the mitochondria and the endoplasmic reticulum, since the group of required oxidative enzymes exists only within these two intracellular organelles. The transport of free cholesterol from the cytoplasm into the mitochondria is the rate‐limiting step. Within the mitochondria, cholesterol is converted to pregnenolone, the immediate precursor, by cytochrome P450 family 11 subfamily A member 1 (CYP11A1). Organs that contain this enzyme, such as the testicles, adrenal cortex, ovaries, and placenta, are the only ones that participate in the synthesis of steroids. The following steps of steroidogenesis are catalyzed by several enzymes that are categorized into two different groups, the cytochrome P450 (CYP) enzymes (oxidative enzymes) and the hydroxysteroid dehydrogenase (HSD) enzymes (Schiffer 2019).
Atorvastatin inhibits the enzyme that converts HMG‐CoA into a cholesterol precursor called mevalonic acid in hepatocytes. Atorvastatin molecules bind reversibly to the active site, changing the conformation of the enzyme, which leads to preventing the enzyme from acquiring a functional structure. Inhibition of HMG–CoA reductase leads to a decline in the intracellular cholesterol, resulting in the activation of a protease that cleaves the sterol regulatory element binding proteins (SREBPs) from the endoplasmic reticulum. SREBPs, in their turn, are translocated in the nucleus and increase the gene expression for LDL receptors in hepatocytes, which leads to a reduction of the circulating LDL and both intermediate‐density lipoprotein (IDL) and very low‐density lipoprotein (VLDL), precursors for LDL (Stancu 2001).
In terms of its potential effect on testosterone, atorvastatin likely decreases the availability of cholesterol, which is a crucial substrate for testosterone production, leading to a decline in serum testosterone. The reduction of androgen levels can be potentially harmful and an adverse side effect in males and females with normal or below normal testosterone levels. However, women with PCOS (hyperandrogenism) may reap benefits from reduced androgen levels. Distinguishing between these two clinical settings is crucial when the effect of statins on androgen levels is being evaluated.
Why it is important to do this review
Based on the mechanism of action (MOA) of atorvastatin, a decline in the levels of testosterone and other androgens will likely occur in patients taking atorvastatin, as atorvastatin decreases the availability of cholesterol, a crucial precursor for androgen production.
Due to the widely increasing use of atorvastatin and the important role of testosterone in both men and women, concerns have been raised about the effect of atorvastatin on serum testosterone in men and women. A non‐Cochrane systematic review attempted to address the effect of statins on testosterone (Schooling 2013). However, this review has a number of limitations. Firstly, it is out of date (the searches range to the end of 2011). It has a limited search strategy, as the review authors searched only MEDLINE and ISI Web of Science, and restricted their search to publications in the English language. Secondly, the review was limited to adults, thus excluding adolescents and children who are potentially particularly vulnerable to reduced serum testosterone levels during their growing years. Thirdly, no protocol was pre‐published, and there are errors in the data reported by the review. Finally, the systematic review was potentially biased, as it was funded by one of the manufacturers of rosuvastatin, MedImmune, LLC. At present, there is no Cochrane Review that addresses the effect of atorvastatin on serum testosterone levels.
Objectives
To quantify the magnitude of the effects of atorvastatin as compared to placebo or no treatment on total testosterone, free testosterone, SHBG, A4, DHEAS concentrations, FAI, and WDAEs.
Methods
Criteria for considering studies for this review
Types of studies
We included all randomized controlled trials (RCTs) with a minimum duration of three weeks, that compared atorvastatin to placebo or no treatment, and measured one or more of the androgenic outcomes.
Types of participants
Participants of both sexes could be of any age, with normal lipid parameters, or any type of hyperlipidaemia or dyslipidaemia. We included participants with various comorbid conditions: diabetes mellitus, hypertension, metabolic syndrome, chronic renal failure, PCOS, or cardiovascular diseases.
Types of interventions
We included atorvastatin doses that ranged from 2.5 mg to 80 mg, administered daily for at least three weeks. We chose a three‐week time window in order to allow for a steady‐state effect of atorvastatin to occur.
Types of outcome measures
Blood concentrations of different androgens.
Primary outcomes
-
The primary outcome of this study was the absolute change in total testosterone concentrations. When the absolute change was reported at different time periods, we chose the absolute change reported at the longest time period.
Secondary outcomes
-
Absolute change in free testosterone
-
Absolute change in FAI
-
Absolute change in DHEAS
-
Absolute change in A4
-
Absolute change in SHBG
(When the absolute change was reported at different periods for these outcomes, we chose the absolute change at the longest period).
-
Patient WDAEs
Search methods for identification of studies
Electronic searches
The Cochrane Hypertension Information Specialist searched the following databases without language, publication year or publication status restrictions:
-
the Cochrane Hypertension Specialised Register via the Cochrane Register of Studies (CRS‐Web) (searched 9 November 2020);
-
the Cochrane Central Register of Controlled Trials (CENTRAL) via the Cochrane Register of Studies (CRS‐Web) (searched 9 November 2020);
-
MEDLINE Ovid (from 1946 onwards), MEDLINE Ovid Epub Ahead of Print, and MEDLINE Ovid In‐Process & Other Non‐Indexed Citations (searched 9 November 2020);
-
Embase Ovid (from 1974 onwards) (searched 9 November 2020);
-
ClinicalTrials.gov (www.clinicaltrials.gov) (searched 9 November 2020);
-
World Health Organization International Clinical Trials Registry Platform (www.who.it.trialsearch) (searched 9 November 2020).
The Information Specialist modeled subject strategies for databases on the search strategy designed for MEDLINE. Where appropriate, these strategies were combined with subject strategy adaptations of the highly sensitive search strategy designed by Cochrane for identifying randomized controlled studies (as described in the Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0, Box 6.4.c (Higgins 2011). The MEDLINE search strategy was translated into the other databases using the appropriate controlled vocabulary, as applicable. We present search strategies for major databases in Appendix 1.
Searching other resources
The Cochrane Hypertension Information Specialist searched the Hypertension Specialised Register segment (which includes searches of MEDLINE, Embase, and Epistemonikos for systematic reviews) to retrieve existing reviews relevant to this Cochrane Review, so that we could scan their reference lists for additional trials. The Specialised Register also includes searches for controlled trials of CAB Abstracts & Global Health, CINAHL, ProQuest Dissertations & Theses and Web of Science.
We checked the bibliographies of included studies and any relevant systematic reviews identified for further references to relevant trials.
Where necessary, we contacted authors of key papers and abstracts to request additional information about their trials, or clarification and further data if trial reports are unclear.
We also searched websites of the following resources, for additional information.
-
Pfizer, Inc ( Pfizer is the maker of Lipitor which is the commercial name of atorvastatin). (www.pfizer.ca/en/our_products)
-
US Food and Drug Administration (www.fda.gov)
-
European Patent Office (worldwide.espacenet.com)
Data collection and analysis
Selection of studies
Two review authors independently, in duplicate, screened the titles and abstracts of records that were retrieved in the search results. We applied the eligibility criteria and used the Covidence software (Covidence) to code these records for possible inclusion as ‘yes,’ ‘no’ or ‘maybe’. We excluded articles if the citation appeared completely irrelevant, after reading the title and abstract. We resolved disagreements in the coding by discussion to reach consensus. We obtained the full‐text papers for all records coded as ‘yes' and 'maybe' at the title and abstract screening stage. Two review authors independently screened the full‐text papers in duplicate to decide which studies met the inclusion criteria. We resolved disagreements by discussion, with adjudication by a third review author (JMW) in the event of non‐consensus. We provided the reasons for exclusion and bibliographic details for all articles excluded after full‐text review. We identified multiple full‐text papers of the same study, which we linked and treated as a single study. We also checked the list of references in the included studies and the articles that cited the included studies in Google Scholar, to identify relevant articles. We created a PRISMA chart diagram to document and report the flow of records and studies through the systematic review process (Liberati 2009).
Data extraction and management
We used an electronic data extraction form based on the Cochrane Public Health template (Data Extraction and Assessment Template), modified to allow extraction of all data required for this review. Two authors independently performed the data extraction, and followed this by a cross‐check of the data. We described the characteristics of each included study in the 'Characteristics of included studies' table.
Our data extraction form included details on study design, type of masking, PICO (i.e. the study's population, intervention, comparison, and outcome) elements, primary and secondary outcomes, effect estimates, mean differences (MDs), confidence intervals (CIs), standard deviation (SD) for each group and main differences, details of interventions and comparators, duration of interventions, baseline characteristics of the participants, statistical analysis, the follow‐up duration, number of participants lost to follow‐up, and funding sources.
We resolved discrepancies by rechecking the data, as well as through discussion and consensus between the two review authors. When we are not able to reach an agreement, a third review author (JMW) acted as arbiter. When necessary, we contacted the authors of the studies to provide clarification on ambiguous information. All extracted data was entered and double‐checked in RevMan 5 software (RevMan 2014).
Assessment of risk of bias in included studies
We independently assessed the risk of bias of the included studies using the Cochrane 'Risk of bias' tool (version 1) following Chapter 8 of the Cochrane Handbook, for the following domains (Higgins 2011):
-
sequence generation (selection bias);
-
allocation sequence concealment (selection bias);
-
blinding of participants (performance bias);
-
blinding of outcome assessors (detection bias);
-
incomplete outcome data (attrition bias);
-
selective outcome reporting (reporting bias); and
-
other potential sources of bias (conflict of interest, funding source)
We produced 'Risk of bias' tables as outlined in the Cochrane Handbook, Chapter 8 (Chapter 8). We reported the information obtained in our bias assessment in the 'Risk of bias' tables associated with each included trial.
We included all studies in the meta‐analysis, regardless of our assessment of the risk of bias. We performed a sensitivity analysis, excluding studies with a high risk of bias, to explore their impact on the treatment effect estimate. This risk of bias across studies informed our assessment of the certainty of evidence, using the GRADE framework.
Measures of treatment effect
For continuous outcomes, we calculated the MD by using mean values, corresponding SDs, and treatment arm sizes. Moreover, we analyzed the data using 95% CIs. We summarized dichotomous data (from studies reporting WDAEs) as risk ratios (RRs) with 95% CIs. We used an intention‐to‐treat (ITT) analysis.
Unit of analysis issues
As we included no cross‐over trials, we had no unit of analysis issues for cross‐over trials. For multi‐arm trials, if a study reported more than one intervention arm (with other drugs than atorvastatin), we identified the relevant intervention arm (atorvastatin in any dose) and included that in the review.
Dealing with missing data
We sought missing or unclear data from reports of included studies by contacting the study authors using the contact information provided in the respective articles.
The most common type of value that was not reported was the SD of the change. If a standard error (SE) was given instead of the SD, we used the formula "SD = SE x square root of n" (with 'n' signifying the number of participants) to calculate the SD. We also calculated SD if the 95% CI, P value, or t value were reported in the included studies, according to Chapter 16 of the Cochrane HandbookChapter 16 . If we were not able to obtain the SD from the study authors or calculate from the values mentioned above, we imputed SD using the following hierarchy (listed from highest to lowest):
-
SD of the end of treatment value;
-
SD of the baseline treatment value; or
-
average weighted standard deviation of the change from other trials in the review (Furukawa 2006).
Assessment of heterogeneity
We assessed statistical heterogeneity in results by inspection of a graphical display of the estimated treatment effects from included studies, along with their 95% CIs, and by formal statistical tests of measures of inconsistency (I2 statistic) (Higgins 2002). If the I2 statistic value was greater than 50%, we investigated the reasons behind the heterogeneity by looking for clinical and methodological differences among trials.
Assessment of reporting biases
As fewer than 10 studies contributed to any meta‐analysis, we did not use a funnel plot to detect the extent of risk of reporting bias, based on the symmetry of the plot, as we had planned to do in the protocol. However, we have discussed the other types of reporting bias, for example multiple publication bias, outcome reporting bias, and language bias (Sterne 2011).
Data synthesis
We used Review Manager 5 (RevMan 2014) to perform data synthesis, and meta‐analyzed the results of clinically and statistically homogeneous studies. We analyzed data for each group (males, females and overall) separately. We performed meta‐analyses using the Mantel‐Haenszel method for dichotomous outcomes (RR) with 95% CIs, using a fixed‐effect model, and the inverse variance method for continuous outcomes as MD with 95% CIs, using a fixed‐effect model.
Subgroup analysis and investigation of heterogeneity
We planned on performing subgroup analysis based on the following.
-
According to various doses
-
Males versus females
-
Age ≥ 60 years versus age < 60 years
-
Children (age < 18 years) versus adult (age ≥ 18 years)
It was not possible to conduct a subgroup analysis based on participant age, as we did not have sufficient data.
Sensitivity analysis
We planned the following sensitivity analyses.
-
To check if the trials with high risk of bias affected the effect estimate of total testosterone, DHEAS, and A4
-
To check the difference between the effect estimates of the outcomes derived by using a fixed‐effect model and a random‐effects model
-
To check the impact of imputed data that we imputed due to missing information on the overall effect size
Summary of findings and assessment of the certainty of the evidence
We used GRADEpro software (GRADEpro GDT 2015) to produce an evidence profile table and used that to generate the 'Summary of Findings' table (Schünemann 2011a; Schünemann 2011b). We considered the following domains in assessing the overall certainty of evidence: limitations in study design and implementation (risk of bias), indirectness of evidence, unexplained heterogeneity or inconsistency of results, imprecision in results and high probability of publication bias. As outlined in the Cochrane Handbook (Schünemann 2011b), we assessed the certainty of the body of evidence for each outcome to be high, moderate, low or very low, and review authors provided comments to support these judgements. RCTs started at a high level of certainty and were downgraded by one level for each domain judged to have some serious concerns (Schünemann 2013).
Results
Description of studies
The summary of the screening process is shown in the PRISMA diagram (Figure 1).
PRISMA diagram for screening studies
Results of the search
Database searching identified 7247 citations. After duplicates were removed, 7225 records remained. After irrelevant citations were removed, 286 records remained. The remaining 286 records were obtained as full‐text articles and assessed for eligibility. Eight citations to six studies (Akbari 2016; Chen 2014; Gokce 2012; Puurunen 2013; Raja‐Khan 2011; Sathyapalan 2009) met the inclusion criteria (Figure 1).
Included studies
We have summarized each included study in the 'Characteristics of included studies'. All six studies were published in English. They included 292 participants who were randomized(154 males and 145 females), but only 265 participants(140 males and 125 females) had their data reported. All of them had a parallel‐group design. Among the six studies included, one was single‐blind, four were double‐blind, and one was open‐label.
Two of the studies included only male participants (N = 140), with a mean age of 68 years. One study (Gokce 2012) included exclusively healthy male participants while the other study (Chen 2014) included male participants with mild dyslipidaemia and osteopaenia. Four of the studies included female participants (N = 125) and the mean age was 32 years. All participants in the four female studies had PCOS. The dose of atorvastatin in the two male studies was 10 mg. The dose of atorvastatin was 20 mg in two of the female studies, and 40 mg in the remaining two studies.
Excluded studies
We excluded 278 articles after reviewing the full‐text articles and recorded the reasons for exclusion. Most articles (262 articles) were excluded because they did not measure and report this review's outcomes of interest. We excluded seven articles because they were not RCTs. In addition, we excluded one study because it measured the testosterone levels but did not report them (see Figure 1 and 'Characteristics of excluded studies').
Risk of bias in included studies
We independently assessed the risk of bias, following the methodology described in Chapter 8 of the Cochrane Handbook (Chapter 8).
We assessed the risk of bias based on seven domains:
-
random sequence generation (selection bias);
-
allocation concealment (selection bias);
-
blinding of participants and personnel (performance bias);
-
blinding of outcome assessment (detection bias);
-
incomplete outcome data (attrition bias);
-
selective reporting (reporting bias);
-
other biases (conflict of interest, industry sponsorship).
We assessed the risk of detection bias and reporting bias in WDAEs separately from their assessment in the other outcomes. We also assessed the risk of attrition bias separately for each of the biomedical outcomes. We classified each domain as being at a low, high or uncertain risk of bias.
In the case of disagreement between review authors, a third review author (JMW) helped to discuss and resolve the disagreement.
The expected direction of risk of bias would be to exaggerate the anti‐androgen effect of atorvastatin as a potential treatment in women with PCOS, and to underestimate the reduction in androgens when it is was studied and measured as a potential adverse effect of atorvastatin treatment in the general patient population (Figure 2; Figure 3).
Risk of bias summary of the included studies
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies
Allocation
We assessed selection bias based on two categories: random sequence generation and allocation concealment.
For random sequence generation, five of the six included studies were classified as having a low risk of bias since they used and described a true randomization method. One study (Akbari 2016) was classified as having an uncertain risk of bias, since the method of randomisation was not described.
For allocation concealment, three included studies (Akbari 2016; Chen 2014; Gokce 2012) were classified as having an uncertain risk of bias because the method of allocation concealment was not reported. The remaining three studies were classified as having a low risk of bias due to the application of appropriate methods, such as central allocation (pharmacy‐controlled randomization and sequentially numbered drug containers) (Puurunen 2013) or allocation by external personnel who were not involved in the trial (Raja‐Khan 2011; Sathyapalan 2009).
Blinding
All six studies were classified as being at low risk of performance and detection bias, as we consider that the individuals measuring laboratory outcomes were not aware of the treatment allocation.
For WDAEs, a lack of blinding could have had an effect. Four studies (Akbari 2016; Puurunen 2013; Raja‐Khan 2011; Sathyapalan 2009) were classified as being at uncertain risk of bias, since the blinding was not described. Two studies (Chen 2014; Gokce 2012) were classified as having high risk of bias, since they were open label and single‐blind, respectively.
Incomplete outcome data
We assessed attrition bias based on six outcomes: total testosterone, free testosterone, free androgen index, A4, SHBG, and DHEAS. For total testosterone, three studies (Akbari 2016; Chen 2014; Sathyapalan 2009) were classified as having low risk of bias given that all participants were included in the final analysis, or missing data were balanced across intervention and control groups, with a similar reason for missing data across groups. Two studies (Gokce 2012; Raja‐Khan 2011) were classified as being at uncertain risk of bias, since 10% of participants did not complete the trial, although the study authors included all participant data in the analysis. One study (Puurunen 2013) was classified as high risk of bias, since 21.1% of participants were not included in the analysis for the atorvastatin group, and 31.6% of participants were not included in the analysis for the placebo group. For A4 and DHEAS, only four studies measured these outcomes. Two studies (Akbari 2016; Sathyapalan 2009) were classified as having a low risk of bias. Puurunen 2013 was classified as high risk of bias, and Raja‐Khan 2011 was assessed to be at uncertain risk of bias. For free testosterone, only one study reported this outcome (Raja‐Khan 2011) and was classified as high risk of bias, since only baseline data was reported. For SHBG and FAI, only two studies measured the outcome: Puurunen 2013 was classified as high risk of bias and Sathyapalan 2009 was classified as low risk of bias.
Selective reporting
We assessed this separately for WDAEs and for all other outcomes. For WDAEs, four studies were classified as having low risk of bias (Akbari 2016; Chen 2014; Gokce 2012; Puurunen 2013) because they recorded and reported the WDAEs in the result section. Two studies (Raja‐Khan 2011; Sathyapalan 2009) were classified as having unclear risk of bias since the WDAEs were not reported separately from withdrawals due to other reasons.
For all other outcomes, three studies (Akbari 2016; Chen 2014; Gokce 2012) were classified as having uncertain risk of bias since no protocol was published and there was insufficient information to evaluate if it was at low or high risk of bias. Two studies (Puurunen 2013; Raja‐Khan 2011) were classified as low risk of bias since all of each study’s prespecified outcomes were reported. One study (Sathyapalan 2009) was classified as having high risk of bias since one reported primary outcome (HS‐CRP) was not prespecified in its trial registration.
Other potential sources of bias
In this domain, the other potential sources of bias assessed included industry sponsorship. Two studies (Chen 2014, Puurunen 2013) were classified as having low risk of bias because they were funded by government grants. Two studies (Akbari 2016; Gokce 2012) were classified as being at uncertain risk of bias since the funding source was not reported. Two studies (Raja‐Khan 2011; Sathyapalan 2009) were classified as high risk of bias because they were funded partially or fully by the Pfizer pharmaceutical company.
Publication bias
We did not pool data from a sufficient number of studies to construct funnel plots for assessing publication bias. However, all studies were published in English. Only six studies from more than 268 placebo‐controlled RCTs identified in our search results reported total testosterone levels, and only two studies reported this outcome in males. There are no studies in healthy females or children (under age 18). This suggests that there are likely RCTs measuring and demonstrating an effect of atorvastatin in reducing testosterone and other androgen levels that were not published, because the results of such RCTs might adversely impact the marketing of atorvastatin.
Effects of interventions
See: Summary of findings 1 Summary of findings; Summary of findings 2 Summary of findings
While in the general patient population, the reduction of testosterone levels caused by the administration of lipid‐reducing interventions including atorvastatin is considered an adverse side effect, this reduction is potentially beneficial in trials in which atorvastatin is administrated as a treatment for PCOS. Because the risk of bias is in the opposite direction in these two clinical settings, we presented and pooled these two clinical settings separately.
1. The potential adverse effect of atorvastatin on androgens
Effect of atorvastatin on total testosterone levels (males)
Based on two studies (Chen 2014; Gokce 2012), in 140 male participants, the mean change in total testosterone, using a fixed‐effect model, was ‐0.20 nmol/L (95% CI ‐0.77 to 0.37; P = 0.49). There was no heterogeneity (I2 = 0%) (Figure 4) .
Forest plot of comparison: Atorvastatin versus control (fixed‐effect model), outcome: Total testosterone in males.
Effect of atorvastatin on free testosterone levels (males)
None of the included studies reported data for this outcome.
Effect of atorvastatin on WDAEs in males and females
Based on six studies (Akbari 2016; Chen 2014; Gokce 2012; Puurunen 2013; Raja‐Khan 2011; Sathyapalan 2009) in 292 participants, the RR for WDAEs using the fixed‐effect model was 1.00 (95% CI 0.14 to 7.37; P = 1.00). There was no heterogeneity (I2 = 0%) (Figure 5).
Forest plot of comparison: 5 Atorvastatin versus control (fixed‐effect model), outcome: 5.7 Withdrawal due to adverse effects (WDAEs).
2. The possible beneficial effects of atorvastatin for the treatment of PCOS
Effect of atorvastatin on total testosterone levels
Based on four studies (Akbari 2016; Puurunen 2013; Raja‐Khan 2011; Sathyapalan 2009) in 125 participants, atorvastatin decreased total testosterone levels. The mean decrease, using a fixed‐effect model, was ‐0.27 nmol/L (95% CI ‐0.50 to ‐0.04; P = 0.02). There was substantial heterogeneity (I2 = 84%) (Figure 6).
Effect of atorvastatin on free testosterone levels
None of the included studies reported data for this outcome. Only one study (Raja‐Khan 2011) reported free testosterone levels before the treatment, but did not report it after the treatment.
Effect of atorvastatin on FAI
Only two studies in females (Puurunen 2013; Sathyapalan 2009) reported FAI in 65 participants. Atorvastatin decreased FAI. The mean decrease, using a fixed‐effect model, was ‐2.59 (95% CI ‐3.62 to ‐1.57; P < 0.00001). There was substantial heterogeneity (I2 = 94%) (Analysis 1.7).
Effect of atorvastatin on SHBG
Only two studies in females (Puurunen 2013; Sathyapalan 2009) reported SHBG in 65 participants. Atorvastatin increased SHBG. The mean increase, using a fixed‐effect model, was 3.11 nmol/L (95% CI 0.23 to 5.99; P < 0.03). There was no heterogeneity (I2 = 0%) (Analysis 1.3).
Effect of atorvastatin on A4
The four studies in females (Akbari 2016; Puurunen 2013; Raja‐Khan 2011; Sathyapalan 2009) reported A4 levels in 125 participants. Atorvastatin decreased A4. The mean decrease, using a fixed‐effect model, was ‐1.37 nmol/L (95% CI ‐2.26 to ‐0.49; P = 0.002). There was substantial heterogeneity (I2 = 68%) (Analysis 1.4).
Effect of atorvastatin on DHEAS
Four studies in females (Akbari 2016; Puurunen 2013; Raja‐Khan 2011; Sathyapalan 2009) reported DHEAS in 125 participants. Atorvastatin decreased DHEAS. The mean decrease, using a fixed‐effect model, was ‐0.63 μmol/L (95% CI ‐1.12 to ‐0.15 P = 0.01). There was no significant heterogeneity (I2 = 9%) (Analysis 1.5).
3. Subgroup analysis and exploring heterogeneity
We planned to conduct several subgroup analyses according to dosage and age. Due to the small number of studies, we were not able to conduct subgroup analyses since it is recommended that there should be at least 10 studies in each of the subgroups.
4.Sensitivity analysis
As planned, we conducted a sensitivity analysis to compare the results using the random‐effects model and the fixed‐effect model. While the reduction in total testosterone, FAI, and androstenedione was statistically significant when the fixed‐effect model was used, the reduction in all outcomes mentioned above was not significant when the random‐effects model was applied (Analysis 2.1; Analysis 2.2; Analysis 2.3). We planned to conduct sensitivity analysis on the studies based on their level of risk of bias. Most of the included studies had a similar risk of bias across all the domains, except for industry sponsorship bias and incomplete data for total testosterone. Due to the inadequate number of studies, we were not able to conduct a sensitivity analysis on the included studies based on industry sponsorship.
Discussion
This review summarizes the effects of atorvastatin on total testosterone, A4, DHEAS, FAI, and SHBG serum levels in males and females. The primary focus of the review was to document the effect of atorvastatin on these outcomes as a potential adverse effect. However, in conducting the review we identified RCTs meeting the inclusion criteria in which patients with PCOS were given atorvastatin with the goal of reducing androgens, as a potential treatment.
For trials studying the possible adverse effect of atorvastatin to decrease androgens, there is a potential bias towards not showing this effect or not reporting RCTs that found this adverse effect. The knowledge that atorvastatin decreased testosterone could result in a negative impact on sales of atorvastatin.
Summary of main results
1. Summary of the results of the possible adverse effect of atorvastatin on androgens
Effect of atorvastatin on total testosterone
In this review, we only found two trials investigating the effects of atorvastatin on androgen levels among males and no studies reporting the effect of atorvastatin on androgens in normal females. This was a surprising and shocking finding, as we anticipated a much larger number of studies looking at the effects of atorvastatin on androgen levels, given the widespread use of atorvastatin. The dearth of studies, the small sample size and the fact that other androgen levels were not reported resulted in the pooled effect of atorvastatin in males showing a statistically non‐significant reduction of total testosterone levels. In addition, the two studies we found used a relatively low dose of atorvastatin, 10 mg/day, when we know that atorvastatin daily doses of 80 mg/day are commonly used. We highly suspect that the companies marketing atorvastatin have successfully suppressed the conduct and/or publication of trials showing the reduction of testosterone and other androgens.
Additionally, it is important to mention that atorvastatin is given as a treatment for children and adolescents with familial hypercholesterolaemia. In this population, where reduction of androgens is more likely to be problematic, we again found no RCTs reporting the effects of atorvastatin on testosterone or other androgens.
Effect of atorvastatin on free testosterone and other androgens
Although free testosterone is a better indicator of the androgen status than total testosterone, no study reported free testosterone levels.
2. Summary of the results of the possible beneficial effects of atorvastatin for the treatment of PCOS
Effect of atorvastatin on total testosterone
For the possible beneficial reduction of total testosterone levels among females with PCOS, the direction of the potential risk of bias is towards a reduction, as the trials are designed to demonstrate that atorvastatin can be used to treat women with PCOS. In this review, we found that atorvastatin decreased total testosterone levels in females by ‐0.27 nmol/l. This finding was based on four studies in females with PCOS.
Effects of atorvastatin on SHBG and FAI
Atorvastatin increased SHBG by 3.11 nmol/l and decreased FAI by 2.59 nmol/L. These results were based on only two studies in females with PCOS. The increase in SHBG levels suggests that more testosterone is bound and that there is less bioavailable free testosterone for tissues. The decrease in FAI corresponds with an increase in SHBG and a decline in total testosterone.
Effect of atorvastatin on DHEAS and androstenedione
Atorvastatin decreased DHEAS concentration in females by 0.63 μmol/L, and androstenedione by 1.37 nmol/L. These results were based on four studies in females with PCOS. These results can be explained by the steroidogenesis pathway; the reduction in DHEAS levels would have led to a reduction in both A4 and testosterone levels. It is important to appreciate that the decline in A4 concentration may be an overestimate, since heterogeneity in the meta‐analysis was substantial and the reduction comes mostly from the Raja‐Khan 2011 study, which was sponsored by the pharmaceutical industry.
Effect of Atorvastatin on WDAEs
In the six trials that reported the impact of atorvastatin on WDAEs, drug therapy had no significant effect on WDAEs compared to placebo or no treatment. However, these were small and short‐term trials and do not represent a good opportunity to determine whether atorvastatin leads to adverse effects.
Overall completeness and applicability of evidence
Surprisingly, there is very little available published data to evaluate the effect of atorvastatin on testosterone and other androgens. Given that we have presented evidence that atorvastatin decreases testosterone and other androgens in women with PCOS, we suspect that atorvastatin also reduces testosterone in other settings. There are likely trials meeting our inclusion criteria, showing this effect of atorvastatin, that have not been published. It is very likely that atorvastatin reduces testosterone as a result of its mechanism of action to reduce cholesterol levels. Further strong evidence that this is occurring is the fact that gynaecomastia is being recognized as an adverse effect of atorvastatin. The two studies in males represented a range in terms of age and health conditions, with a mean age of 68 years. It is possible that atorvastatin does decrease androgen levels more significantly in a younger population .
The participant populations were older males with normal libido (Gokce 2012) and elder men with mild dyslipidaemia (Chen 2014). The overall evidence generated in this review is insufficient to answer the question in the populations studied, and cannot be extrapolated to men and women of all ages, or to adults with sexual dysfunction.
Regarding the studies in females, the participant population was women with PCOS, predominantly in premenopausal age (younger than age 40). Unfortunately, we do not have any trials studying the effect of atorvastatin on females at any age without PCOS. It is, however, likely that atorvastatin does decrease androgen levels in women without PCOS, and that this could have adverse consequences.
There was substantial heterogeneity in the meta‐analysis of the effect of atorvastatin on testosterone in females with PCOS, due to the presence of one study (Sathyapalan 2009). This study was funded by the Pfizer pharmaceutical company. The mean reduction of total testosterone was greater in this study compared to other studies. Thus, we suspect that this reduction may be an overestimate.
There was substantial heterogeneity in the meta‐analysis of the effect of atorvastatin on androstenedione in females, due to the presence of one study (Raja‐Khan 2011). This study was funded by the pharmaceutical industry. The mean reduction of total testosterone was greater in this study, compared to other studies. The unbalanced baseline between the comparison groups in Puurunen 2013 may explain the substantial heterogeneity between the two studies in the FAI meta‐analysis.
Quality of the evidence
We graded the overall certainty of evidence using the GRADE approach, with the GRADEpro GDT software (GRADEpro GDT 2015), and formulated 'Summary of findings' tables, as outlined in the protocol.
We constructed two 'Summary of findings' tables to show the certainty of evidence and a summary of the effects on the outcomes of interest in females (total testosterone, A4, DHEAS, FAI, SHBG, and WDAEs), and total testosterone in males.
In this review, the certainty of evidence ranged from low to very low, which suggests that the estimated effect of atorvastatin may be substantially different from the true effect. Regarding the effect of atorvastatin on total testosterone levels in males, the certainty of evidence was downgraded to very low due to the small size of the studies, wide confidence intervals, and the high risk of bias. Regarding the effect of atorvastatin on total testosterone, A4, and FAI in females, the certainty of evidence was downgraded to very low due to the small size of the studies, significant inconsistency, and the high risk of bias in some domains.
The certainty of evidence for DHEAS and SHBG in studies in females was downgraded to low due to the small sample size and the high risk of bias in some of the studies.
See summary of findings Table 1 and summary of findings Table 2.
Potential biases in the review process
We employed a robust methodology to assess biases and used comprehensive search strategies to minimize potential for bias by the authors in the selection process of included studies. A few amendments have been implemented on the protocol which were explicitly mentioned and justified.
We have identified several limitations within this review, which are related to the studies that were included. We highly suspect that there are more studies which meet the inclusion criteria and were not published (publication bias), despite our efforts to search the grey literature. Our review's included studies and its data are only applicable to specific population groups, and therefore should not be generalized. The overall sample size of the studies was not sufficiently large to answer our research question. Finally, the reporting of outcomes in some of the included studies tended to be poor in nature. The reporting of the SD of the change in levels was the main problem. In one of the studies in males (Gokce 2012), the authors did not report the SD of the change or the SD of the means, either before or after the intervention. Thus, we imputed the SD of the change from other studies, using the guidance of Furukawa (Furukawa 2006).
In the studies in females, the SD of change was missing, and we had to calculate it from the reported P values. In some studies (Akbari 2016; Puurunen 2013) we imputed the SD of change using the SD of final values. Accordingly, the results should be interpreted with caution. Another major problem related to data reporting was the units of measurements. In one of the included studies (Gokce 2012), the units were reported incorrectly (the testosterone unit was reported as ng/ml instead of ng/dl). Additionally, units were essentially missing, and the author did not report reasons for not including units (Akbari 2016). Furthermore, withdrawal due to adverse effects was not explicitly reported in some of the studies.
Agreements and disagreements with other studies or reviews
One published review (Schooling 2013) examined the effect of atorvastatin on total testosterone. As discussed earlier in this review, Schooling 2013 pooled the effect of all statin agents on testosterone, by assuming that they all had similar effects. We limited the focus of this review only to atorvastatin. Schooling 2013 concluded that statins reduce testosterone in both males and females. However, there are a number of limitations to the Schooling 2013 review, which we have listed in the background.
Another published review (Yang 2019) examined the effect of atorvastatin on DHEAS. Yang pooled the effect of all statin agents on DHEAS and conducted a subgroup that analyzed the effects of atorvastatin and simvastatin separately. According to Yang 2019, atorvastatin significantly reduced DHEAS, compared to simvastatin. The conclusion of Yang 2019, which was based on three of the studies that were included in this review (Puurunen 2013; Raja‐Khan 2011; Sathyapalan 2009), aligns with our review's conclusion.
Statin‐induced gynaecomastia was reported by several studies and case reports as a potential adverse effect among male patients (Jeong 2019; Roberto 2012; Skeldon 2018). Although our review did not find significant reduction in androgens among male patients taking statins, this could be due to publication bias and the limitations of studies included.
This review did not detect a significant difference in WDAEs between atorvastatin and placebo, since only two studies reported WDAEs. A previous review (Adams 2015) with 43 short‐term RCTs reporting WDAEs, also showed no effect of atorvastatin on WDAEs.
PRISMA diagram for screening studies
Risk of bias summary of the included studies
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies
Forest plot of comparison: Atorvastatin versus control (fixed‐effect model), outcome: Total testosterone in males.
Forest plot of comparison: 5 Atorvastatin versus control (fixed‐effect model), outcome: 5.7 Withdrawal due to adverse effects (WDAEs).
Comparison 1: Atorvastatin versus control (fixed‐effect model), Outcome 1: Total testosterone in males
Comparison 1: Atorvastatin versus control (fixed‐effect model), Outcome 2: Total testosterone in females
Comparison 1: Atorvastatin versus control (fixed‐effect model), Outcome 3: Sex hormone binding globulin (SHBG)
Comparison 1: Atorvastatin versus control (fixed‐effect model), Outcome 4: Androstenedione
Comparison 1: Atorvastatin versus control (fixed‐effect model), Outcome 5: Dehydroepiandrosterone sulfate (DHEAS)
Comparison 1: Atorvastatin versus control (fixed‐effect model), Outcome 6: Withdrawal due to adverse effects (WDAEs)
Comparison 1: Atorvastatin versus control (fixed‐effect model), Outcome 7: Free androgen index (FAI)
Comparison 2: Atorvastatin vs control (random effects model), Outcome 1: Total testosterone in females
Comparison 2: Atorvastatin vs control (random effects model), Outcome 2: Androstenedione
Comparison 2: Atorvastatin vs control (random effects model), Outcome 3: Free androgen index (FAI)
| Population: Male participants Intervention: Atorvastatin Comparison: Placebo or no treatment | |||
| Outcomes | Number of participants (RCTs) | Mean difference of atorvastatin compared to placebo (95% CI) | The certainty of evidence (GRADE) |
| Total testosterone | 140 (2) | ‐0.20 nmol/L (‐0.77 to 0.37) | Very low 1, 2, 3 |
| Acronyms and grades | |||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; RCT: randomized controlled trial | |||
| GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect | |||
| 1 Unclear risk of selection bias. 2 95% confidence interval around the best effect estimate includes both a reduction and an increase in total testosterone. 3 Evidence was derived from only two studies, which had several limitations. | |||
| Population: Adult female participants with PCOS Intervention: Atorvastatin Comparison: Placebo | |||
| Outcomes | Number of participants (RCTs) | Mean difference of atorvastatin compared to placebo (95% CI) | The certainty of evidence (GRADE) |
| Total testosterone | 125 (4) | ‐0.27 nmol/L (‐0.50 to ‐0.04) | Very low 1, 2,3 |
| FAI | 65 (2) | ‐2.59 (‐3.62 to ‐1.57) | Very low 1, 2, 3 |
| SHBG | 65 (2) | 3.11 nmol/L (0.23 to 5.99) | Low 1, 3 |
| Androstenedione | 125 (4) | ‐1.37 nmol/L (‐2.26 to ‐0.49) | Very low 1, 2, 3 |
| DHEAS | 125 (4) | ‐0.63 μmol/L (‐1.12 to ‐0.15) | Low 1, 3 |
| Acronyms and grades | |||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; DHEAS: dehydroepiandrosterone sulphate; FAI: free androgen index; RCT: randomized controlled trial; SHBG: sex hormone binding globin | |||
| GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect | |||
| 1 High risk of bias in source of funding and sponsorship bias in more than one study. 2 Significant heterogeneity. 3 Small sample size. | |||
| Tests of blood: normal findings | ||
|---|---|---|
| Free testosterone pg/mL | ||
| Age/Tanner stage | Male | Female |
| Postmenopausal | 0.6‐3.8 | |
| 7 months‐9 years (Tanner stage I) | ≤ 3.7 | < 2.2 |
| 10‐13 years (Tanner stage II) | 0.3‐21 | 0.4‐4.5 |
| 14‐15 years (Tanner stage III) | 1.0‐98 | 1.3‐7.5 |
| 16‐17 years (Tanner stage IV) | 35.0‐196 | 1.1‐15.5 |
| 18‐19 years (Tanner stage V) | 41.0‐239 | 0.8‐9.2 |
| Free testosterone % | ||
| Adult male | 1.6‐2.9 | |
| Adult female | 0.1‐0.3 | |
| Total testosterone ng/dL | ||
| Male | Female | |
| 7 months‐9 years (Tanner stage I) | < 30 | < 30 |
| 10‐13 years (Tanner stage II) | < 300 | < 40 |
| 14‐15 years (Tanner stage III) | 170‐540 | < 60 |
| 16‐19 years (Tanner stage IV, V) | 250‐910 | < 70 |
| 20 years and over | 280‐1080 | < 70 |
| Dihydrotestosterone | ||
| Adult male | 240‐650 pg/mL | |
| Adult female | ≤ 300 pg/mL | |
| Total testosterone and free testosterone levels by sex and age (Pagana 2015). | ||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1.1 Total testosterone in males Show forest plot | 2 | 140 | Mean Difference (IV, Fixed, 95% CI) | ‐0.20 [‐0.77, 0.37] |
| 1.2 Total testosterone in females Show forest plot | 4 | 125 | Mean Difference (IV, Fixed, 95% CI) | ‐0.27 [‐0.50, ‐0.04] |
| 1.3 Sex hormone binding globulin (SHBG) Show forest plot | 2 | 65 | Mean Difference (IV, Fixed, 95% CI) | 3.11 [0.23, 5.99] |
| 1.4 Androstenedione Show forest plot | 4 | 125 | Mean Difference (IV, Fixed, 95% CI) | ‐1.37 [‐2.26, ‐0.49] |
| 1.5 Dehydroepiandrosterone sulfate (DHEAS) Show forest plot | 4 | 125 | Mean Difference (IV, Fixed, 95% CI) | ‐0.63 [‐1.12, ‐0.15] |
| 1.6 Withdrawal due to adverse effects (WDAEs) Show forest plot | 6 | 292 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.00 [0.14, 7.37] |
| 1.7 Free androgen index (FAI) Show forest plot | 2 | 65 | Mean Difference (IV, Fixed, 95% CI) | ‐2.59 [‐3.62, ‐1.57] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 2.1 Total testosterone in females Show forest plot | 4 | 125 | Mean Difference (IV, Random, 95% CI) | ‐0.27 [‐0.87, 0.34] |
| 2.2 Androstenedione Show forest plot | 4 | 125 | Mean Difference (IV, Random, 95% CI) | ‐1.56 [‐3.32, 0.20] |
| 2.3 Free androgen index (FAI) Show forest plot | 2 | 65 | Mean Difference (IV, Random, 95% CI) | ‐1.89 [‐6.30, 2.51] |
