Lessons from the Past: Avoiding Placebo Generated Increased Hair Counts.

INTRODUCTION

During the past 30 years, the evaluation of scalp hair and in particular, methods employed in the Food and Drug Administration (FDA)-approved clinical trials failed to fully appreciate potential confounding issues. In 1985,[1] hair regrowth was reported to show the reversal of male pattern hair loss in subjects receiving topical 2% minoxidil and placebo, as evaluated by visually counting hair in a 1” diameter circle (2.5 cm). Concerns were raised following this publication as to how a placebo could induce hair regrowth in male pattern hair loss.[2] As part of an Upjohn sponsored, double-blind, 2% minoxidil, placebo-controlled study, visual hair counting was compared with the validated quantitative unit area trichogram.[3] In the placebo group, no significant increase in any hair variable was observed with the unit area trichogram analysis, while significant increases were recorded in the same site and same individual, with visual hair counting. This study showed counting experience resulted in increased hair counts where none existed. This study resulted in visual hair counting being abandoned, and the development of photographic imaging techniques ensued.

The vast majority of hair researchers then adopted an FDA approved photographic method in double-blind placebo-controlled hair growth evaluation studies, which relied on “visible hair” count measurements to determine treatment efficacy. However, these investigations also reported increased hair growth in subjects treated with placebo lotions, again raising questions as to how “visible hair” on photographs could be interpreted as improved hair growth in an individual using a placebo lotion?

In 2011, the evidence-based (S3) guidelines (revised 2017) for the treatment of androgenetic alopecia in men and women was published[4] and acknowledged “that the reported placebo rate in most of the minoxidil studies is very high.” In 2012, a Cochrane review of androgenetic alopecia[5] was also concerned about placebo generated hair growth. However, both publications failed to address either the nature of this confounding or its implications on the observed drug-associated therapeutic effect.

Data reviewed

We evaluated exclusively those studies rated by independent experts as being of the highest quality from the publications in the S3 guidelines and examined the nature of the so-called “placebo effect” in relation to the reported active drug effect limited to a 1-year duration. For each study, we selected the placebo data time point reflecting the longest exposure to placebo, between 24 and 48 weeks (mean 41 weeks ± 9 standard deviation [SD]) and compared this value where possible with the reported baseline hair count.

There were 13 placebo-controlled studies involving a total of 4743 participants. One thousand four hundred and ninety-one participants were included in six studies where topical minoxidil was compared with a placebo lotion, both of which contained propylene glycol. About 74% of participants in this group were female. Three thousand two hundred and fifty-two male participants were included in seven oral finasteride or dutasteride studies, compared with men taking placebo pills.

Active drug dosages matched those in clinical practice, i.e., topical application of 2%, and 5% minoxidil solutions (1 ml twice daily) or oral finasteride (1 mg) or dutasteride (0.5/0.1 mg) daily. The mean treatment duration was similar between oral and topical groups. All studies employed the FDA approved photographic method and hair count changes from baseline are given as hair per cm2(absolute) or per cent (relative) changes from baseline [Table 1 and Figure 1].

Table 1

Hair counts in placebo treated subjects

S3*	Study	Placebo Subjects	Gender	Weeks	Topical placebo, study type	Absolute change, hair/cm2	Relative change (%)
50	Olsen 1991	30	Female	32	Minoxidil 2	20.6	13.4
53	Whiting 1992	33	Female	32	Minoxidil 2	17.0	10.6
48	Jacobs 1993	346	Female	32	Minoxidil 2	19.1	13.7
47	De Villez 1994	308	Female	32	Minoxidil 2	9.9	7.1
33	Olsen 2002	393	Male	48	Minoxidil 5	3.9	2.6
49	Lucky 2004	381	Female	48	Minoxidil 5	9.4	6.8
	Total	1491	Mean±SD	37±8	Mean±SD	13±6.6	9±4.3
S3*	Study	Placebo subjects	Gender	Weeks	Oral placebo, study type	Absolute change, hair/cm2	Relative change(%)
57	Kaufman 1998	1553	Male	48	Finasteride 1 mg	−4.1	−2.7
60	Leyden 1999	326	Male	48	Finasteride 1 mg	−2	−0.9
64	Roberts 1999	693	Male	48	Finasteride 1 mg	−3.9	−2.1
65	Stough 2002	18	Male	48	Finasteride 1 mg	−4	−2.5
67	Van Neste 2001	212	Male	48	Finasteride 1 mg	−10.1	−5.2
61	Olsen 2006	416	Male	24	Dutasteride 0.1 mg	−6.3	−3.5
73	Stough 2007	34	Male	48	Dutasteride 0.5 mg	−3.8	¥
	Total	3252	Mean±SD	45±9	Mean±SD	−5±2.6	−3±1.4
Topical placebo versus oral placebo P <0.0001

Data were taken from 13 double-blind-placebo-controlled clinical trials rated of the highest quality in the S3 report.[4] The columns give the S3 reference#, first author and year of publication (study), number of subjects, gender, duration of study (weeks), and administration route. The changes in hair count from baseline as absolute hair per cm2 and whenever possible, the relative percentage values. The mean and SD of changes are shown after topical or oral placebo at the bottom two lines for each administration route with their corresponding P values. *S3 study Blumeyer et al.,[4] ¥No baseline data given. SD – Standard deviation

Figure 1

Hair count changes from baseline are displayed as hair per cm2 (absolute) or per cent (relative) changes from baseline in a series of clinical studies. (a) The mean absolute and relative hair count changes from baseline are shown for active (plain black bars) versus placebo groups (empty or hatched bars). Active groups included topical 2%–5% minoxidil + propylene glycol (PG M 2%–5%), 5% minoxidil foam (Foam M 5%), or daily oral finasteride 1 mg or dutasteride 0.5 or 0.1 mg (Oral active). The vehicle in topical placebo groups contained the same concentration propylene glycol (PG placebo empty bars) as the active topical lotions, foam (without PG) in one single study (foam placebo; oblique hatched bars), and oral placebo (vertical hatched bars). Placebo source data appear in table while the mean drug associated changes were taken from the tables published in the S3 report.[4] (b) Displays net mean hair count either as the absolute or relative change. The “net effect” was obtained by deducting the absolute or relative changes reported in placebo groups from those given in active groups for topical lotions (speckled bar, PG M 2%-5% minoxidil), foam (empty bar, Foam M 5% minoxidil), and oral preparations (vertical hatched bar; Oral Active)

Statistical analysis

Absolute changes in hair counts and percentage change versus baseline in the placebo groups are detailed in Table 1, while drug effects are given in Figure 1. Statistical levels of significance (P ≤ 0.05) for mean changes in hair counts were evaluated employing Student's t-test.

RESULTS

Table 1 presents the absolute and relative changes (mean ± SD) of “non-vellus” hair counts reported for placebo groups. All studies employing the twice daily, propylene glycol-containing placebo-containing lotion, reported increased hair counts compared to baseline. In contrast, all oral placebo studies reported a decrease in hair counts compared to baseline. The absolute and relative (%) mean hair counts values, between the two groups (topical placebo versus oral placebo) was statistically significant (P<0.0001).

Figure 1a presents the absolute and relative (%) hair counts for the active and placebo compounds, while Figure 1b presents the net differences, i.e., active minus placebo values.

CONCLUSIONS

From the published placebo-controlled studies employing an FDA approved photographic method, we have highlighted that there is probably technological confounding for the claimed increases in hair counts, raising concerns about the extent of the claimed topical minoxidil efficacy. We suggest that the reported “hair regrowth” associated with topical placebo lotions might be due to propylene glycol influences. A position supported by a mean decrease in the oral placebo group (−3%) and only a negligible increase (+3.4%) in the placebo arm of a topical minoxidil study, that did not contain propylene glycol.[6] Furthermore, the use of non-FDA approved hair evaluation methods, for example, in vivo microscopy[7] and the unit area trichogram[3] also found no significant increases in placebo hair counts in subjects treated with propylene glycol-containing lotion.

We note that while the Expert Reviewers of the S3 and Cochrane publications raised concerns about the reported increase in placebo hair counts they offered neither explanation for this phenomenon nor was the implication of this confounder considered. How such an important observation might have been generated, and its impact on the efficacy of the active treatment was impossible to address in this communication. Identifying potential confounders and their influence on hair counts cannot be ignored in future hair growth efficacy studies. We would suggest before undertaking clinical trials where hair evaluation is an outcome that the evaluation methods be fully validated before the start of the investigation.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.