Paperonium

← Back to the plain-language case

Abstract says vitamin D3 rose in Scribble knockout mice; Results and Fig. 3E show a fall

Proc Natl Acad Sci U S A (2023) · doi:10.1073/pnas.2220851120 · PMC10266016

Summary

The abstract states that knocking out Scribble in renal proximal tubules "caused significant plasma phosphate elevation and increased aggregate vitamin D3 levels". The paper defines a single aggregate vitamin D3 measure: three metabolites, each normalized to littermate controls and then combined. Fig. 3E is the only place it is plotted. The Results text and the Fig. 3 legend both describe that measure as decreased, and the points plotted in Fig. 3E agree: digitized, the six tamoxifen-induced knockout values have a mean of 63.2% of control, and 5 of the 6 lie below 100%. Vitamin D is one of the two in vivo findings the abstract reports, so a reader of the abstract alone gets the direction of that finding reversed.

What the paper reports

Abstract (as archived in PMC, after the Significance statement):

The loss of Scribble impacted serum phosphate and vitamin D levels and caused significant plasma phosphate elevation and increased aggregate vitamin D3 levels, whereas blood glucose levels remained unchanged.

Results, section "Loss of Scribble Impacts Renal Phosphate Homeostasis and Serum Vitamin D Levels", second paragraph:

Combined 1,25(OH)2D3 + 24,25(OH)2D3 + 24(OH)D3 levels, as indicated in Methods, decreased only in Tam-injected animals and not in corresponding Cre littermates (Fig. 3E).

Fig. 3 legend:

(B) Targeted Scribble KO mice display increased serum phosphate (C), normal levels of calcium and glucose (D), decreased 1,25[OH]2-vitamin D (1,25[OH]2D) levels (E), and normal levels of PTH and FGF23 (F and G).

Materials and Methods, "Mice.", where the aggregate is defined:

Vitamin D3 was measured as described (52) on littermate pairs. 1,25 vitamin D3, 24,25 vitamin D3, and 25 vitamin D3 forms were normalized to control levels and combined.

Fig. 3E (an image; the axis title and group labels were read from the archived figure file pnas.2220851120fig03.jpg): the y-axis is "Plasma Vitamin D % control", the groups are Veh, Tam and Cre, and a bracket marks Veh vs Tam with "*" and Veh vs Cre with "ns".

Statements elsewhere that give no direction. Significance statement:

The loss of this interaction impacts serum phosphate and vitamin D levels in Scribble KO mice.

Discussion, second paragraph:

Targeted loss of Scribble in proximal tubules resulted in pronounced alterations to serum phosphate and vitamin D levels, whereas glucose, calcium, PTH and FGF23 levels remained unaffected.

Discussion, third paragraph, on mechanism:

PTHR activates the 25-hydroxyvitamin D3-1α-hydroxylase catalyzing the formation of biologically active 1,25(OH)2-vitamin D3.

What the paper's own numbers show

Method. The text gives no vitamin D values, means or test statistics, so the direction can be checked in two ways: by comparing what the text says in each place, and by reading the plotted points in Fig. 3E. For the second, I took the figure as archived in PMC and Europe PMC (708 × 927 px). I calibrated the y-axis from its five tick marks (0, 50, 100, 150 and 200%; largest residual 0.5 percentage points) and located each dot as the centroid of its connected block of dark pixels. As a check, the six Veh (control) dots, which are fixed at 100% by the normalization, come out at 99.8%. The mean computed from the digitized dots matches the mean bar drawn in the figure to within 0.7 percentage points for Tam and exactly for Cre. The paper does not say which test produced the "*". Because each knockout value is already a ratio to its own littermate control, a one-sample test against 100% is the natural analogue, and I report it only to show that the plotted points are consistent with the asterisk. Every number below is printed by recompute.py.txt.

Quantity Location Reported Recomputed from the paper's own figure
Direction of the aggregate vitamin D3 measure Abstract "increased" decrease (opposite sign)
Direction of the same measure Results, second paragraph "decreased only in Tam-injected animals" decrease (agrees)
Direction of panel E Fig. 3 legend "decreased … levels (E)" decrease (agrees)
Control (Veh) dots Fig. 3E fixed at 100% 99.8% (calibration check)
Tam (knockout) dots, % of control Fig. 3E plotted only 109.2, 78.0, 55.8, 49.2, 44.0, 42.7
Tam mean (median) Fig. 3E mean bar drawn, no value given 63.2% (52.5%); drawn bar reads 62.5%
Change vs control Fig. 3E — −36.8 percentage points
Tam dots below 100% Fig. 3E — 5 of 6
Tam 95% CI of the mean (t, df = 5) Fig. 3E — 35.9% to 90.4%; excludes 100%
Test of Tam vs control Fig. 3E "*" one-sample t = −3.47, two-sided p = 0.018; exact Wilcoxon p = 0.0625
Cre-only littermates, % of control Fig. 3E "ns" 150.8, 133.4, 126.0, 109.9, 78.5, 69.2; mean 111.3% (95% CI 77.8 to 144.9; p = 0.43)
Digitizing error needed to move the Tam mean to 100% — — 21.9 px for every dot; a dot is about 5 px across

The direction does not depend on how precisely the dots were located: every Tam dot would have to be misplaced by about four dot diameters for the mean to reach 100%. The only group whose mean lies above 100% is Cre, and that difference is marked "ns" and is not attributed to loss of Scribble.

Figure

The abstract says the aggregate vitamin D measure increased; the Results text, the Fig. 3 legend and the points plotted in Fig. 3E all show a decrease

Figure 1. Top: the two text statements about the same aggregate vitamin D measure. The abstract (orange outline) says "increased"; the Results text and Fig. 3 legend (grey outline) say "decreased". Bottom: the six knockout (Tam) values read from Fig. 3E, as % of littermate control (open blue circles), with their mean of 63.2% (filled blue circle) and its t-based 95% interval of 35.9% to 90.4% (blue line); the vertical line marks the control level of 100%. Blue values are recomputed by recompute.py.txt from the paper's own figure; they are not values printed in the paper. The axis starts at zero.

Design candidates for review. recompute.py.txt also writes five candidate layouts of this figure, fig1_A to fig1_E (fig1 is currently design D). Only the chosen design will be kept as fig1.

Why it matters

Explanations considered

Status

Suggested correction

In the abstract, replace

caused significant plasma phosphate elevation and increased aggregate vitamin D3 levels

with

caused significant plasma phosphate elevation and decreased aggregate vitamin D3 levels

so that the abstract agrees with the Results text, the Fig. 3 legend and the data in Fig. 3E. The authors may also wish to make the third metabolite the same in Results ("24(OH)D3") and Methods ("25 vitamin D3"), to describe panel E in the legend as the combined measure, and to give the test used for Fig. 3E with its n and P value.