Risk prediction model for probability of PHN after HZ | JPR

Introduction

Herpes zoster (HZ) is characterized by a neurocutaneous inflammatory reaction induced by latent varicella-zoster virus (VZV) reactivation within the host. It appears as a vesicular rash with a dermatomal distribution.¹ Postherpetic neuralgia (PHN) is the most prevalent and refractory complication of HZ that persists for over 3 months after the occurrence of shingles. Affected individuals may experience enduring and intense pain for numerous years, or even decades, which significantly affects their daily life and increases the economic burden.²

HZ has a global incidence of 3–5/1000 person-years, with 5–35% of patients progressing to PHN.³ PHN represents a intricate neuropathic pain disorder, and its underlying mechanisms remain enigmatic. Present therapeutic modalities for PHN predominantly encompass oral analgesics, nerve blocks, pulsed radiofrequency, and spinal cord stimulation.⁴ Nevertheless, most of the therapeutic effects remain unsatisfactory.⁵

Considering the profound effect of PHN on patient health and quality of life, the anticipation of PHN occurrence in patients with acute HZ has become the focal point of contemporary research, which promises preventive strategies for those with HZ. According to several studies, older age, severe acute pain, and an intense rash are key risk factors for PHN.^6,7 Additionally, a limited number of studies have suggested that gender, immunosuppression, diabetes, systemic lupus erythematosus, and respiratory diseases (asthma, COPD, obstructive sleep apnea, and lung cancer) are linked to PHN.^8,9 However, prevailing studies have often confined themselves to clinical factors rather than incorporating serological or virological tests, which has potentially led to inaccurate outcomes.

Inflammation is acknowledged as a pivotal facet of chronic pain.¹⁰ Emerging research speculates that heightened peripheral blood inflammation contributes to chronic pain.^11,12 Indicators of peripheral blood inflammation, such as neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), lymphocyte-to-monocyte ratio (LMR), C-reactive protein-to-albumin ratio (CAR), and systemic immune inflammation index (SII), are used for the diagnosis, treatment, follow-up management, and prognosis prediction of various diseases, including cancer, acute pulmonary embolism, diabetic foot-induced sepsis, and coronary artery disease. This is because of the accessibility, reproducibility, non-invasiveness, and cost-effectiveness of these indicators.^13–17 Furthermore, some studies have examined the correlation between indicators of peripheral blood inflammation and chronic pain, such as postsurgical pain, low back pain, fibromyalgia, and neuropathic pain, in diabetes.^11,18–20 However, studies exploring the role of these indicators in the context of HZ are limited.

The models currently employed for predicting PHN risk after acute HZ have limitations. Therefore, we here retrospectively analyzed 209 HZ patients to establish and validate a novel clinical prognostic model by incorporating clinical factors and markers of peripheral blood inflammation (NLR, PLR, LMR, CAR, and SII) for predicting PHN development.

Materials and Methods

Patient Population

The electronic health records of 209 patients diagnosed as having HZ during the acute phase were extracted from the archives spanning January 2019 to June 2023 at Jinling Hospital, Nanjing University School of Medicine, People’s Republic of China. The study was approved by the medical ethics committee (2019NZKY-030-02), and explicit informed consent forms were procured from all patients. Our study complies with the principles of the Declaration of Helsinki (October 2013).

The study inclusion criteria were as follows: (1) diagnoses of HZ and PHN were established based on clinical symptoms and supplementary examinations; (2) within a 14-day timeframe from onset; and (3) comprehensive clinical records were available. The exclusion criteria were as follows: (1) absence of 3-month follow-up data; (2) presence of severe organ dysfunction, including the heart, lung, or kidney; and (3) recent occurrence of acute severe infection.

Data Collection and Definition

Clinicopathological data were systematically extracted from the patient’s electronic medical records, including age, gender, body mass index (BMI), distributional side of the rash, Numeric Rating Scale (NRS-11) score, hypertension, and diabetes mellitus. The assessment of pain intensity was conducted utilizing the NRS scale, which ranges from 0 to 10, with a delineation of severe pain corresponding to a score of ≥7 on the NRS.

A peripheral blood count was performed before any treatment modality was initiated. The indices NLR, PLR, LMR, CAR, and SII were calculated and documented using the prescribed formulas: NLR = neutrophil/lymphocyte, PLR = platelet/lymphocyte, LMR = lymphocyte/monocyte, CAR = C-reactive protein/albumin, and SII = platelet × neutrophil/lymphocyte. The last interview was conducted in July 2023.

Statistical Analysis

Continuous variables were analyzed using t-tests, while categorical variables were analyzed using chi-square tests. Univariate and multivariate logistic regression analyses were conducted to determine independent predictors of PHN in patients with acute HZ, with odds ratio (OR) and corresponding 95% confidence intervals (CI) employed as effect estimates.

Based on the results of multivariate logistic regression analyses, we established the risk prediction model and nomogram model. Receiver operating characteristic (ROC) curves were plotted, and the area under the curve (AUC) was calculated using the pROC package. The calibration of our model was assessed using the Hosmer–Lemeshow goodness-of-fit test and calibration curve.

Moreover, the clinical utility of the nomogram was validated by conducting the decision curve analysis (DCA) to quantify the net benefit under different threshold probabilities. The model was internally validated using the repeated sampling method (Bootstrap method). Statistical analyses were conducted using R software (version 4.3.2), with statistical significance set at p

Results

Baseline Characteristics of the Study Population

In total, 215 patients diagnosed as having acute phase HZ were included in this study, and of them, 6 patients were excluded (Figure 1). Finally, 209 patients who met the inclusion criteria were analyzed in this study. Of the total patients enrolled, 62 patients (30%) developed PHN. All patients had received antiviral drug therapy before the baseline assessment. Among the participants, 130 (62%) and 75 (36%) patients had male gender and were aged ≥60 years, respectively. The patients with HZ were stratified into two groups based on the presence or absence of PHN. These patients revealed statistically significant differences in age, hypertension, NRS-11 score, PLR, and LMR (P Table 1 presents the demographic and clinical characteristics of the two groups.

Figure 1 Flowchart depicting the study protocol.

Univariate and Multivariate Logistic Analyses

Logistic regression analysis was performed to ascertain factors contributing to PHN development. According to the univariate regression analysis, PHN occurrence in individuals was notably correlated with HZ and age (p = 0.010), hypertension (p = 0.012), acute phase NRS-11 score (p = 0.002), PLR (p p = 0.037). The five significant risk factors were then included in the multivariate logistic regression analyses. The results of this analysis disclosed that age (OR = 2.309, 95% CI: 1.163–4.660, p = 0.018), acute phase NRS-11 score (OR = 2.837, 95% CI: 1.294–6.275, p = 0.009), and PLR (OR = 1.015, 95% CI: 1.010–1.022, p Table 2).

Table 2 Univariate and Multivariate Logistic Regression Analyses

Construction of a Nomogram Model for Predicting PHN After Acute HZ

As depicted in Figure 2, we here formulated a nomogram model after the multivariate logistic regression analyses to predict PHN occurrence from HZ. This model relies on the three identified independent predictors. A high total score indicated a high PHN probability. For example, an HZ patient aged ≥60 years and with a score of ≥7 on the NRS-11, PLR of 245, and has an estimated probability of PHN of 86%.

Figure 2 Representatives of nomogram prediction model (A) and dynamic nomogram (B).

Predictive Model Validation

Subsequently, we conducted the ROC curve analysis to estimate the discriminative performance of our predictive model. The AUC was 0.7871 (95% CI: 0.6342–0.8155) and thus indicated commendable diagnostic performance (Figure 3A). The Bootstrap method was applied for internal validation, which resulted in an AUC of 0.7758 (Figure 3B). The calibration curve of our predictive model exhibited that the predicted and observed probabilities were highly consistent (Figure 4), thereby confirming the model’s excellent calibration. Additionally, DCA was performed to evaluate the clinical application value. According to our DCA plot, the nomogram holds substantial clinical value, thus offering remarkable benefits to patients (Figure 5).

Figure 3 Receiver operating characteristic curve (ROC) for the discrimination of the predictive model (A) and internal verification model (B).

Figure 4 Calibration curves of the predictive PHN risk nomogram.

Figure 5 Decision curve analysis (DCA) of the predictive PHN risk nomogram.

Discussion

A recent meta-analysis reported that the incidence of HZ for all age groups is 4.28 cases per 1000 population-years, with the estimated HZ case numbers exceeding 6 million in China.21 Among these cases, 12.6% of patients are expected to develop PHN. Once PHN ensues, patients may experience severe and intractable pain of diverse nature, such as burning, throbbing, cutting, or tearing, which detrimentally affects their quality of life and imposes a substantial economic burden on them, their families, and their society at large.³ Therefore, a clinically practical assessment tool that accurately predicts PHN development must be explored, thereby guiding prevention and intervention strategies during acute phase HZ to mitigate PHN incidence. However, predicting the emergence of PHN is a formidable challenge.

In our study, univariate logistic regression analyses were performed to identify potential risk factors for PHN. Five variables, namely age, hypertension, acute phase NRS-11 score, PLR, and SII, were statistically significant (P

Older age is a risk factor known to influence concurrent PHN. The incidence of HZ and PHN markedly increases with age, particularly after the age of 60 years.^7,22,23 The overall PHN proportion among HZ patients aged ≥60 years is approximately 20.83%.²⁴ In our study, the incidence of PHN among HZ patients aged ≥60 years was 29.33%, whereas that among patients aged 5,25 This age-related increase in disease is closely related to the weakening of VZV-specific T cell mediated immunity and elevated virus levels after zoster reactivation.^26–28

Pain is the predominant clinical manifestation of HZ. The pain intensity during the acute phase is positively correlated with the severity of nerve injury, which directly influences PHN incidence.²⁹ In this study, PHN incidence significantly differed between the high- and low-NRS-11 groups (50% vs 25%, P 30,31 Fujiwara et al further examined the association between pain catastrophizing (Pain Catastrophizing Scale score ≥ 30) in the acute phase of HZ and PHN development, but no significant correlation was noted.³²

PHN primarily manifests as neuropathic pain, and its exact pathogenesis remains complex. Immune-mediated inflammation plays a crucial role in neuropathic pain development by generating nerve cell-damaging pro-inflammatory cytokines and molecules, thereby leading to persistent pain.³³ The intricate balance between pro- and anti-inflammatory cytokines regulates neuropathic pain development.³⁴

Several new composite parameters reflecting inflammatory indices including NLR, PLR, LMR, CAR, and SII have recently been used to evaluate the body’s inflammatory response. These parameters have been considered novel, accurate inflammatory biomarkers in many chronic or acute inflammation-related diseases.^35,36 In the present study, these inflammatory indices, along with traditional clinicopathological variables were subjected to univariate and multivariate analyses, ultimately revealing PLR as an independent risk factor for PHN. This is the first study identifying PLR as an inflammatory marker in HZ and underscoring its utility in PHN prediction. An elevated PLR indicates a relatively decreased lymphocyte, whereas increased platelet counts. Lymphocytes reflect cell-mediated adaptive immunity.³⁷ T lymphocytes are the primary components of lymphocytes and account for 65–75% of the total lymphocytes in peripheral blood.^38,39 In the HZ pathogenesis, the body’s defense chiefly relies on cellular immune responses, and the T lymphocyte’s function is pivotal in maintaining normal immune function.^40,41 T lymphocytes decreased and immune function was impaired in HZ patients,^42,43 which accelerated VZV reactivation and resulted in neuropathic pain. Platelets, a typical blood cell component, are produced by mature megakaryocytes in bone marrow responsible for the systemic inflammatory response and immune regulation in the body.^44,45

Other possible predictors of PHN were also evaluated. Underlying conditions including diabetes and cardiovascular diseases serve as an additional zoster severity-related burden.⁴⁶ In the present study, diabetes and hypertension were included in the analysis. The two groups exhibited significant differences in terms of hypertension, although hypertension was excluded from the following multivariate logistic regression analyses. The role of sex in PHN remains controversial, with published results being conflicting. Some studies have suggested a higher risk in female patients than in male patients, whereas some found no association with sex.⁶ In the present study, sex was not significantly associated with PHN development. These conflicting results may warrant further investigation.

Limitations

Despite its many strengths, this study has some limitations. For instance, the primary constraint lies in the relatively small sample size and limited clinical data of HZ (n = 209), which potentially introduces bias to the results. In addition, to mitigate these limitations and enhance the robustness of the findings, further polycentric prospective large-scale studies are warranted to comprehensively investigate the influencing factors of PHN in patients with HZ.

Conclusions

Collectively, our findings suggest that age, acute phase NRS-11 Score, and PLR are independent predictors of PHN in patients with HZ.