BGPT: Paper Review: Association of left ventricular myocardial dysfunction with diabetic polyneuropathy

Explore by Goal

Quick Explanation Copied

Key finding (from the paper)

In 166 adults with diabetes and preserved LVEF (≥55%), lower global longitudinal strain (GLS) was independently associated with presence of diabetic polyneuropathy (DPN) (multivariable OR for GLS: 0.55, p<0.001), and tibial F-wave latency was the strongest NCS determinant of DPN (OR 1.43, p<0.001), with a significant negative correlation between GLS and F-wave latency (r = −0.43, p<0.001).

All statistics and effect directions above are reported by the authors.

Paper DOI:

Long Explanation

Paper Review (Evidence-weighted, Skeptical): Association of left ventricular myocardial dysfunction with diabetic polyneuropathy

Retrospective cross-sectional human study (LVEF preserved; DPN by nerve conduction study; LV systolic mechanics by STE-derived GLS).

DOI: 10.1007/s10396-018-0898-6

1) What the authors actually tested

Association question: Are markers of DPN severity associated with LV longitudinal systolic dysfunction in patients with diabetes who have preserved LVEF?
Operational definitions: LV longitudinal systolic function = GLS from 2D speckle-tracking echocardiography (STE). DPN severity = NCS Baba Classification (Grades 0–IV) using lower-extremity nerves (tibial F-wave latency, MCV/CMAP, sural SCV/SNAP, peroneal MCV/CMAP).

2) Visuals first: group-level quantitative patterns reported

Note: the paper provides group means/SDs for several variables; the figures below reflect those reported values.

GLS means/SDs plotted directly from the paper’s Table of baseline comparisons (DPN vs non-DPN).

Tibial F-wave latency means/SDs plotted from the paper’s NCS comparison table.

3) Statistical associations reported (and how to interpret them skeptically)

3.1 Multivariable determinants of DPN

GLS independently associated with DPN in stepwise forward multivariable logistic regression: OR 0.55, p<0.001.
Within NCS features, multivariate analysis identified tibial F-wave latency as the most closely related NCS parameter to DPN: OR 1.43, p<0.001.

3.2 Cross-variable correlation (GLS ↔ F-wave latency)

The paper reports a negative correlation between GLS and tibial F-wave latency: r = −0.43, p<0.001.

3.3 Severity pattern (does it track linearly with grade?)

GLS was significantly lower in Grade I or higher DPN vs Grade 0, but the paper reports no significant differences among Grades I–IV.

Skeptical interpretation constraints

Cross-sectional design: association at one timepoint cannot establish temporal order (whether LV dysfunction precedes DPN, follows it, or both share upstream pathology).
Exclusion not equivalent to complete ascertainment: coronary artery disease was excluded using medical records and laboratory findings, without coronary angiography/CT angiography—leaving residual possibility of occult coronary disease contributing to strain abnormalities.
Population & generalizability: NCS Baba classification standard values may differ by ethnicity; study involved Japanese subjects only.

4) Measurement reliability: reproducibility evidence inside the paper

The authors report ICCs for intra-/inter-observer agreement for strain metrics.

ICCs extracted from the paper’s reproducibility section: GLS intra 0.95 (95% CI 0.88–0.98), GLS inter 0.93 (0.82–0.97); GCS intra 0.97 (0.93–0.99), GCS inter 0.92 (0.86–0.95); GRS intra 0.93 (0.85–0.96), GRS inter 0.90 (0.77–0.94).

5) What the findings might mean biologically (and what remains unknown)

Known from the paper

The authors position GLS reduction as early/subclinical LV longitudinal dysfunction in diabetes that co-occurs with early DPN (Grade I).
The paper’s association pattern is stronger for presence of DPN (non-DPN vs DPN; Grade 0 vs Grade I+) than for monotonicity across DPN grades I–IV.

Uncertainty / blind spots

Residual confounding: even with stepwise multivariable logistic regression, diabetic complications (e.g., kidney disease, microvascular burden) and treatment differences can create confounding structures not fully captured—especially given differential prevalence of retinopathy/nephropathy in the DPN group.
Measurement coupling risk: GLS and diastolic indices (E/A, e’, E/e’) are derived from the same echocardiographic session, and NCS severity is also time-adjacent (within 1 week). Shared pre-analytic factors (hydration status, autonomic tone, medication timing) can inflate associations without proving shared pathophysiology. This coupling is plausible given study design (explicitly reported temporal proximity), but the paper does not quantify such bias.
Standardization and external validity: strain quantification is vendor/software and protocol dependent; the authors used a GE EchoPAC workflow and reported ICCs, but external inter-lab reproducibility is not established here.

6) Practical takeaways (strictly from the paper’s evidence scope)

In diabetic patients with preserved LVEF, STE-derived GLS is associated with DPN presence and also correlates with a key NCS marker (tibial F-wave latency).
The association appears most evident at the transition from no DPN (Grade 0) to mild DPN (Grade I+), not as a strongly graded monotonic dose-response across Grades I–IV (as analyzed by the authors).
The paper itself calls for prospective work to determine whether GLS predicts cardiac events/diabetic microangiopathy progression.

7) Links to author-level reviews (browsable in BGPT)

Feedback:

Updated: April 04, 2026