A paper published last month in PLOS One by a group of investigators from the University of Verona in Italy states that Arnica montana Stimulates Extracellular Matrix Gene Expression in a Macrophage Cell Line Differentiated to Wound-Healing Phenotype. Given my abiding interest in pharmacognosy and ethnobotany, I was suitably intrigued because the extract derived from Arnica montana, a European flowering plant of the sunflower family, is likely to be biologically active due to the presence of certain sesquiterpene lactones (same class of substances as present in the plant-derived anti-malarial Artemisinin), the plant metabolite flavonoids (substances with some in vitro anti-oxidant and anti-inflammatory activity), and derivatives of thymol (phenolic substance with antimicrobial action). Like many bioactive phytopharmaceutical substances, Helenalins, the sesquiterpene lactones and their fatty acyl esters in Arnica montana, are toxic in high concentrations, but have anti-inflammatory properties via its inhibition of the transcription factor NFkB.

How did I hear about Arnica? It is a popular herbal medicine, often topically applied as ointment/cream, for pain and inflammation, and there appears to be some clinical evidence that topical Arnica may be as effective as ibuprofen, a non-steroidal anti-inflammatory agent, in mitigating pain from osteoarthritis. And of course, courtesy my homeopathy-aficionado family members, “Arnica mont” has long been a familiar name. Therefore, it was with quite some academic interest that I delved into the PLOS One paper.

The Experimental Setup

Macrophages play an important regenerative role in injured tissue, which led the authors to hypothesize these blood cells as a possible target of Arnica’s action; to test in vitro, they used a human monocyte/macrophage cell line (THP-1) in an environment that pushes these cells towards wound healing and tissue remodeling functions, exposed the cells to Arnica extracts, and estimated the changes in gene expressions via RNA content of the cells. In an earlier study (whose details are not available to me because of a closed-access publication), they had estimated the expressions of certain immune-response associated genes; in the present one, they ambitiously expanded their target to include the whole human transcriptome (RNAs made from the whole human genome) using a high-throughput next-generation sequencing technique called RNA-seq. As an interesting aside, even though most of the genome is transcribed into RNA, it appears that majority of the RNAs are “non-coding”, that is, not translated into proteins – which often makes the analysis of RNA-seq data challenging. However, the authors have adroitly addressed this challenge by using a functional annotation system to focus only on the sequences of known “protein encoding” genes – which I thought was a rather neat stratagem. Additionally, the authors used a macrophage-based in vitro wound-healing model suitable for studying the cell–matrix and cellular interactions necessary for wound healing in order to investigate the bioactivity of Arnica extracts.

Methodological Concerns

The main treatment modality, an ethanol-extract of Arnica montana, was supplied to the investigators by the study’s funding agency, the French Homeopathy manufacturer, Boiron, as a 1C (1:100) dilution in 3 part to 7 part ethanol to water (30% ethanol/water), and the authors prepared all the subsequent serial study dilutions in pure water, such as 2C (1 part in 10,000 parts), 3C (1:1,000,000; 6X in decimal notation), 5C (1:10 billion; 10X), 9C (1:1018; 18X) and 15C (1:1030; 30X). My regular readers (Oh that rare breed!) would immediately recognize the last as a dilution in which, following laws of Physics, there cannot be even a single molecule of the original Arnica’s substance left. The Control was equally diluted starting with 30% ethanol water, serially diluted in pure water for the corresponding number of times.

The authors have kindly provided (unusual for most homeopathy papers) a quantitative idea of the substances in the extract —albeit only for a single component, the sesquiterpene lactones, and not others. Apparently, sesquiterpene lactone content of the original ethanolic extract (the ‘Mother tincture’) was 360 micro-gram per milliliter, approximately equivalent to 10−3 Mol/L, or 1 millimolar (mM) solution. So, the study dilutions of 2, 3, 5, 9 and 15C were, respectively, 10 microMolar (µM), 100 nanomolar (nM), 10 picoMolar (pM), 1 zeptoMolar (zM), and beyond calculation solutions. What this means is that beyond the 9C dilution, the dose is chemically equivalent to the Control. Interestingly, for most studies with replicate samples —as proper in experiments— the dilution used was 2C, which certainly contains measurable quantities of the active principles in the Arnica extract. For the higher dilutions, the tests were done on pooled samples for each dilution; this is problematic, because pooling of samples abrogates observations of biological variabilities, which is an important lacuna from an analytical standpoint.

Treatment effect (Arnica vs. control) was measured from the RNA-seq data, by calculating the logarithm base-2 (Log2) of the ratio between the expression unit (RPKM) of each gene in Arnica-treated samples and Control-treated samples (Log2 Fold Change). Genes with Log2 Fold Change values that were significantly positive (“up-regulated”) or negative (“down-regulated”) were defined as differentially expressed genes (DEGs). This merits a bit of understanding: for each gene, Arnica-related RPKM and Control-related RPKM are, respectively, the numerator and denominator of a ratio. If the ratio is 1 (whose Log2 is zero, ‘0’), it indicates ‘No Change’, that the gene has same expression level under both; as the ratio increases from 1, the numerator RPKM increases relative to the denominator RPKM, indicating Arnica is associated with higher gene expression compared to the control, whereas if the reverse is seen, the ratio falls below 1, indicating Arnica’s association with lower expression for those particular genes compared to that of the Control. Consequently, the Log2 value is, respectively, positive or negative. Two concepts are rather important in this context:

  1. The logarithmic value Log2, by its mathematical nature, amplifies the differences in the ratios; for instance, the Log2 values of 0.1, 0.2 and 0.3 actually represent much smaller Fold Change ratios of, respectively, 1.07, 1.15 and 1.23; similarly, negative Log2 values of -0.1, -0.2 and -0.3 represent 0.93, 0.87 and 0.81 Fold Change ratios. Whether these small numerical values at all have any biological significance must be tested rigorously using functional means and assessed with strict statistical criteria. I am not sure the authors have taken this principle into consideration during their analysis (See below).
  2. The dichotomy of the Log2 values (positive vs. negative) makes it tempting to assign functional significance, as the authors have done – indicating a positive value to represent up-regulation of gene expression, and negative, down-regulation. Down-regulation in this context would essentially mean that the treatment, Arnica, is suppressing the gene expression by some mechanism. However, it is important to remember that especially for genes with constitutively low expressions, a lower-than-control expression value (leading to a negative Log2) may simply mean that the treatment is ineffective in the context – as opposed to active suppression of expression. Why this is important will be clearer in the discussion on analytical issues (See below).

Analytical Issues

I am not going to comment on the part of the study with 2C dilutions of the Arnica extract. Biological effects are not implausible at that high concentration, and may well fall under the purview of biochemical pharmacognostic research to understand this plants medicinal properties better. Instead, I would deconstruct some of the issues in the analysis undertaken by the authors especially for the higher dilution and pooled samples. I must acknowledge that the authors’ provision of Supplemental Data at the PLOS One website has been instrumental in helping me understand their study.

Taking the example of two random genes from Table S2 (the actual names and functions are mentioned in the Supplemental table, but not important for this illustration), check the distribution of RPKMs for Arnica (AM) and Control (CTRL) in case of each experimental replicate, as well as the mean of replicates and compare it with the pooled sample RPKM. For the higher dilution analyses, the RPKMs from the pooled samples have been arbitrarily compared with the mean RPKM of the Control samples. The distribution of AM and CTRL values and their relative distance for each replicate are both subject to biological and experimental variations which would likely affect the outcome of the analysis, and yet pooling the samples disregards the effects of such variations, making the conclusions less robust.

RPKM values gleaned from Table S2

Values graphically represented from S2 Table: Expression values (RPKM) and differential expression (Log2 Fold Change) of a series of extracellular matrix genes selected from the Reactome database. (Click to embiggen.)

The same applies to the data presented in the paper and Supplemental Table S1. Again, another crucial factor, this time considering the higher dilution data. I have already mentioned about the small Fold Change values; normalized by a logarithmic transformation, these values may reach statistical significance (notwithstanding the small sample size which would lower the power), but their biological significance in aggregate is highly doubtful. This becomes clearer if we look at the Fold Change data, illustrated here using data from Table S1.

Table S1 - Fold Change

Graphs drawn with data from Table S1: Expression values (RPKM) of Control and Arnica-treated cells and differential expression (Log2 Fold Change) of the series of genes reported in Table 1. (Click to embiggen.)

Look closely at both the top and the bottom panel. The top panel represents genes whose Fold Changes are above 1, which means exposure to Arnica dilutions has increased their gene expressions. Or has it? If you follow the potentization (succussion followed by dilution) principle of Classical Homeopathy, you’d expect progressively greater effect for each gene with each successive dilution. But no, there is no such pattern, and for the most part the fold changes in the 9C and 15C dilutions hover around 1. Interestingly, in the lower panel, where the fold changes are largely below 1 (which means the Control RPKMs were higher than Arnica RPKMs), there appears to be a much more pronounced gradation effect, with the 15C dilutions all reaching close to the value of 1 (that is, no change vs. the control). This is a significant outcome which doesn’t seem to have been adequately discussed by the authors.

Wound Healing Assay

This assay is a well-done elegant assay supported by the plausibility of biological activity of Arnica at a high concentration, 2C. Therefore, I have not much to say here, except: (a) I had expected the author to comment at greater detail on the fact that in absence of IL-4, the wound-healing function of Arnica appears to be completely negated, the mechanism of which merits investigation; (b) given that the cells had to be pre-incubated for 24h in Arnica and Control solutions prior to wounding, I hope that this assay was done in a blinded manner, whereby the person doing the wounding and/or making subsequent measurements would be unaware of the Arnica/Control status of each sample — this is extremely important to avoid investigator bias; and (c) the leap from the observed modest in vitro effects to speculations about in vivo efficacy of Arnica, made in the discussion, seems unwarranted and unsupportable by evidence.

In Summary…

So did Arnica extract, at a reasonably high concentration, do all that was claimed in the paper? It did, yes, in terms of changing the expression of certain extracellular matrix associated genes, but only within the narrow context of macrophages with switched-on (via IL-4) wound healing phenotype; within the same context, it showed some quantifiable in vitro wound-healing functions. Given the conditions in (a) and (b) above, translation of Arnica efficacy in vivo is still a stretch – pending empirical evidence. The authors shifted the onus of function on the substance itself by this comment:

…given the variety of Arnica m. effects and the multiplicity of its alkaloids, flavonoids, and sesquiterpene lactones, it is conceivable that the picture of its action is much more complex and could involve modulation of different cells and further pathways. The field of pharmacologic regulation of connective tissue and cell matrix by natural and chemical compounds is open to further studies and developments.

This is indeed true. Funnily enough, this also flies directly in the face of the much-vaunted ‘individualized’ or ‘holistic’ therapy claims that homeopaths often make. The rather poor and inchoate outcome of the high dilution (including the implausible homeopathic 15C) study within this paper’s investigation also casts serious doubts on homeopathic Arnica’s purported efficacy. And finally, a point that I must make: I have no problem with Boiron’s providing funding to study the properties of the source of their homeopathic preparation. But legions of homeopathy aficionados on Twitter and other social media venues (many of whom have been lionizing this study as evidence for homeopathic Arnica’s benefits) ought to take a pause the next time they find a pet conspiracy theory involving real medicine and ‘Big Pharma’.

Paper discussed:

Marzotto, M., Bonafini, C., Olioso, D., Baruzzi, A., Bettinetti, L., Di Leva, F., Galbiati, E., & Bellavite, P. (2016). Arnica montana Stimulates Extracellular Matrix Gene Expression in a Macrophage Cell Line Differentiated to Wound-Healing Phenotype PLOS ONE, 11 (11) DOI: 10.1371/journal.pone.0166340