An analysis of lactate dehydrogenase (LDH) levels in advanced stage IV melanoma of the skin: prognostic capabilities and demographic variability

Despite its limitations, in part due to decreased specificity in advanced disease, serum lactate dehydrogenase (LDH) is one of few serum factors used in cancer staging. This study quantifies the predictive capabilities of LDH in stage IV melanoma of the skin and explores the validity of suggested demographic discrepancies which may exist in its use. The 1975–2017 Surveillance Epidemiology and End Results (SEER) database was queried for stage IV cutaneous melanoma cases. Demographic characteristics were compared between LDH groups using chi-square and t tests. Subsequent Cox multivariable regression was performed to assess survival differences. 334 cases of stage IV cutaneous melanoma (average age: 63.0 years) with measured serum LDH levels were identified. Of these patients, 150 (44.9%) had normal LDH, 112 (33.5%) had LDH < 1.5 × upper limit of normal (ULN), 57 (17.1%) had LDH 1.5–10 × ULN, and 15 (4.5%) had LDH > 10 × ULN. Lower incomes were associated with higher LDH; individuals with incomes < $50,000 had the greatest proportion of LDH 10 × ULN (19.2%; p = 0.0031). LDH > 10 × ULN also had the lowest proportion of White patients (p = 0.04). On Cox multivariable survival analysis, increasing LDH levels showed increased risk of death (LDH < 1.5 × ULN: HR = 2.05, p = 0.01; LDH 1.5–10 × ULN: HR = 1.46, p < 0.001; LDH > 10 × ULN: HR = 5.91, p < 0.001). This study reaffirms the utility of LDH as a significant predictor of mortality with incremental severity, suggesting possible use for mortality projections. We note that Black patients and those with lower incomes may be more likely to have an elevated LDH. Older age groups and presence of ulceration among patients with stage IV melanoma were also associated with a greater risk of mortality.


Introduction
Malignant melanoma is a common cancer with a low 5-year survival rate of 16.7% for stage IV tumors [1,2]. Treatment of melanoma has great economic implications, costing each stage IV melanoma patient between $34,103 and $152,244 annually, and the US melanoma patient population 65 years and older $390 million cumulatively every year [3,4]. Thankfully, the majority of the 106,110 new cutaneous cases of skin melanoma were detected prior reaching stage IV [3,4].
Previous studies have suggested metastasis and serum lactate dehydrogenase (sLDH) levels as possible predictors of survival, with sLDH remaining as one of the few serum factors used for cancer staging today [5,6]. However, the utility of sLDH as a prognostic factor, specifically for metastasis, has yet to be performed on a national scale [7]. Significant limitations with current use of sLDH as a biomarker have been noted, such as LDH not being an actively secreted enzyme and so it is only released by active cell death, occurring more frequently in malignant neoplasms, and the number of false-positives with hemolysis, hepatocellular and muscular diseases, and other malignancies [8,9]. While the specificity of sLDH level as a biomarker for detecting presence of melanoma increases with stage, up to 92% for stage IV, sensitivity of sLDH decreases as stage progresses with stage IV having 79% sensitivity, leading to a propensity for false negatives [10]. Although the use of sLDH level have been used in the American Joint Committee on Cancer (AJCC) staging system for melanoma for over a decade, demographic characterization of sLDH and delineating which patients benefit most from its predictive capabilities remain unexplored on a national level [11]. Furthermore, LDH's ability to predict various responses to treatment, such as to anti-PD1/PD-L1 therapy, remains an area of interest [12]. The largest disparity in survival rates between Black and White patients in absolute terms, as estimated by the American Cancer Society in its 2021 statistical data, was found to be present in melanoma (25%) [13]. Furthermore, with shorter survival times due to melanoma in Blacks and other non-White races, it is possible that predictive tools often used in staging for melanoma, like sLDH, lack similar effectiveness in predicting mortality for people of color compared to White patients. This study aims to use the statistical power of a large national database to analyze the effects of sLDH levels on perioperative outcomes, as well as the associations of sLDH with demographic characteristics in patients with stage IV melanoma.

Materials and methods
A retrospective analysis of the 1975-2017 Surveillance Epidemiology and End Results (SEER) database was queried for stage IV skin melanoma cases. Data were collected in compliance with the ethics and terms in the SEER Research Data Use Agreement. We obtained several clinical characteristics for each patient including age, sLDH, primary site of tumor, and disease-specific survival. We also collected other demographic information like insurance status and income level.
Once stage IV skin melanoma cases were identified, eligibility requirements for inclusion into the study included presence of measured sLDH within the SEER database. Patients with an unknown sLDH were excluded. SLDH was stratified into four groups based upon pre-existing SEER categories: normal, sLDH < 1.5 × upper limit of normal (ULN), sLDH 1.5-10 × ULN, and sLDH 10 × normal. These levels included only pre-treatment sLDH levels. Breslow thickness was divided at the limit of 4 mm, to allow for comparisons with previously published studies [14,15].
Univariable and multivariable analyses were performed using Microsoft Excel (Microsoft, Seattle, WA) and SAS Software (SAS Studio Release 3.8, Cary, North Carolina). Demographic characteristics of the included patients were compared using a combination of chi-squared and ANOVA tests between sLDH groups. Finally, Cox multivariable survival analysis was performed on investigated variables using disease-specific survival. Statistical tests used a two-sided α value of 0.05.

Results
Of 194,491 patients with melanoma of the skin in the SEER database, 1523 patients with stage IV skin melanoma were selected. Of these patients, 334 patients were found to have stage IV melanoma with complete sLDH information and were included in the final study. Of these patients, 150 ( Upon comparison of demographic variables between sLDH groups, ( Table 1) there were no differences in age (p = 0.997) or gender (p = 0.5643) between the four sLDH groups. Significant differences were observed in race, with the highest proportion of Black patients (13.3%) falling in the sLDH > 10 × ULN group (p = 0.0359, Table 1). Significant differences were also observed in terms of income; patients in the sLDH > 10 × ULN group had the greatest proportion of individuals with incomes < $50,000 (33.3%) whereas patients in the normal sLDH group had the highest proportion of individuals with incomes > $65,000 (69.3%) (p = 0.0031, Table 1).
Of specified primary tumor sites, the most common site among all sLDH groups was skin of the trunk. There were no significant differences in primary tumor location between the sLDH groups (p = 0.9165, Table 1). Additionally, no differences were observed in tumor ulceration (p = 0.2577), number of tumors (p = 0.5200), and Breslow thickness of tumors (p = 0.7939) between the sLDH groups.

Discussion
In this study, we elucidate the relationship between sLDH levels and various demographic factors as well as examine the contributions of demographic, disease etiology, and biomarker variables to mortality risk in stage IV melanoma patients. Between 1975 and 2017, 334 patients were found with stage IV melanoma and known sLDH levels in the SEER database. Using SEER-generated categories of normal sLDH, normal to 1.5 times upper limit of normal (ULN), 1.5-10 times ULN, and greater than 10 times ULN, we examined the association between sLDH and perioperative variables, including demographic characteristics and mortality. We found that 44.9% of these 334 patients had normal sLDH levels, 33.5% had normal to 1.5 times ULN, 17.1% had 1.5-10 times ULN, and 4.5% had greater than 10 times ULN, with 78.1% of the original stage IV melanoma population having unknown sLDH values. The distribution of patients in each sLDH level is similar to the distribution expected based on previous studies. Although previous studies have explored the effects of sLDH on postoperative outcomes in patients with melanoma, few studies have investigated the relationship between sLDH and postoperative outcomes in stage IV melanoma patients specifically using a large national database [16,17].
Although previous studies on racial discrepancies in sLDH levels are limited, our findings indicate an association between race and sLDH levels, with Black patients having a higher proportion of members with sLDH greater than 10 times ULN (p = 0.04, Table 1) compared to White patients. While the exact cause for this difference is unknown, one possibility is that healthcare disparities lead to Black patients being diagnosed at a later stage and less likely to receive chemotherapy, radiation therapy, and surgery in a timely manner, resulting in greater tumor growth and sLDH levels [18,19]. In addition, Black patients have a higher percentage of acrolentigious melanomas (including subungual melanomas), associated with a much worse prognosis than typical cutaneous melanomas on sun-exposed skin [20]. Subungual melanoma represents approximately 1-3% of all cutaneous melanomas in the White population, yet up to 75% of Black melanomas [21]. Therefore, delay in diagnosis and treatment for Black patients may also be due to the rarity of melanoma in Blacks overall and the anatomic sites of its appearance especially on the palms and soles, compared to more commonly sun-exposed regions in White patients [22,23]. Following various forms of melanoma treatment and controlling for differences in treatment quality and socioeconomic status, Black patients were still found to have poorer survival, indicating that non-treatment-related factors like tumor biology and cancer biomarkers may be possible explanations for racial disparities in survival [24]. For example, sLDH was found to be positively associated with fibulin-1 and ROS in Blacks only; this increase in extracellular matrix remodeling and oxidative stress may have implications for melanoma prognosis [25].
Our Cox survival analysis found no difference in mortality risk of any race subpopulation compared to that of White patients when controlling for sex, income, primary tumor site, ulceration status, number of tumors, and sLDH level. This finding is contrary to the conclusions from other studies that Black and Hispanic patients have significantly higher mortality rates compared to white patients [26,27]. The insignificant racial difference in mortality risk may be due to the limited sample size of Black patients with stage IV melanoma documented in the SEER database. Furthermore, as a number of variables were controlled for in our model compared to those of other studies solely examining the effect of race on mortality, it is likely that some of the mortality differences by race can be explained by these other variables. The prognosis of stage IV melanoma is already poor with limited options for treatment compared to earlier stages of melanoma, so there is less possibility for race-based variations in mortality risk [28,29]. Ward-Peterson et al.'s study [30] analyzed SEER data over a 30-year period and found that while Black melanoma patients had a higher cause-specific and all-cause risk of mortality compared to white patients, this difference became insignificant after adjusting for site and stage of diagnosis. These findings emphasize the need for early detection and treatment of melanoma, which can be accomplished by improving physician efforts to clinically and histologically diagnose melanoma in black patients, as well as educate the black population on melanoma [31].
We also noted an association between income and sLDH levels, with patients in the lowest income class of less than $50,000 having a higher proportion of sLDH levels greater than 10 times ULN, a relationship that has not been explored in the literature. However, our Cox survival analysis found that income level, like race, did not impact risk of mortality when controlling for confounders, contrary to income being inversely related to risk of mortality in other studies [32]. The difference between our findings and those in other studies could be due to our use of the levels of income stratified by the SEER database that are more granular than those used in these other studies. Patients of lower socioeconomic status are more likely to rely on Medicaid or lack insurance altogether, while patients of higher socioeconomic status are more likely to have commercial insurance. It has been well documented that melanoma patients with commercial insurance have better survival outcomes compared to those who are covered by Medicaid or lack insurance [33,34]. Higher sLDH levels in these lower-income patients provides a possible explanation for their poorer survival outcomes but since SEER only provides insurance status data for 2007 onwards, we cannot correlate insurance status with survival using our sample.
Our finding that patients with sLDH levels normal to 1.5 times ULN, 1.5-10 times ULN, and greater than 10 times ULN have worse survival compared to patients with normal sLDH levels appears to be consistent with the Warburg Effect as well as clinical findings on the use of sLDH as a prognostic factor in melanoma [35][36][37]. We also found that ulceration is associated with an increased risk of mortality as in accordance with other studies [38,39]. Ulceration has been found to be a predictor of mortality independent of age, sex, tumor thickness, site, and mitoses per millimeter, while sLDH has been found to be a predictor of mortality independent of gender, visceral organ involvement, platelet count, age, performance status, tumor thickness, and time to stage IV diagnosis [40][41][42][43][44]. However, ulceration and sLDH have not been shown in the literature to be predictors of mortality independent of each other. Presence of ulceration indicates a more aggressive tumor, and the prognostic ability of sLDH independent of ulceration demonstrates that sLDH has predictive value in melanoma cases beyond reflecting the aggressiveness of the tumor.
All older age groups of patients had a greater risk of mortality compared to the group of patients less than 50 years old except for the group of patients between 60 and 69 years, who did not have a significantly different risk of mortality. There was no significant difference in sLDH levels between age groups, affirming that age and sLDH levels are independent predictors of mortality in melanoma patients [45,46]. The findings of other studies are mixed on the effect of age on mortality. Lideikaitė et al. [47] found that patients aged 65 years and older have a lower 10-year survival rate than that of younger cohorts, but there is no significant difference in melanoma-specific mortality among these cohorts. Chao et al. [48] and Lasithiotakis et al. [49] found that as age increases, the incidence of poor prognostic factors like Breslow thickness, incidence of ulceration, and male gender increases. Some studies have found that age does not correlate with risk of mortality in melanoma patients at all, while others contend that age correlates positively with risk of mortality in localized and regional disease but not in distant disease [50,51]. Given the ambivalent conclusions regarding the association between age and mortality, it is important that future studies using large databases calculate stagespecific and melanoma-specific mortality, and control for localized or regional disease versus distant disease.
There are limitations to consider in this study. First, since our study is based on the SEER database, variations in ICD-9 and -10 coding and incomplete information upon entry of patient values by health practitioners caused many of the sLDH values and ulceration statuses of patients to be recorded as unknown. Therefore, despite the use of a very large sample with Stage IV melanoma, the vast majority of these patients had to be excluded due to missing information for these important prognostic factors. Another factor contributing to missing values for variables that require patient follow-up is that patient movement out of the 20 geographic areas covered by SEER could disproportionately affect certain patient subpopulations more than others. Many melanoma cases are reported to cancer registries like SEER only years after diagnosis, which may impact mortality statistics because as melanoma treatment and care improves in more recent years, mortality decreases [52]. Although SEER is a national database, the incidence of melanoma in minority populations is sufficiently low such that the sample size of Black patients makes it difficult to draw conclusions with absolute certainty. Furthermore, given the diversity in a condition like melanoma, and various demographics affected more by certain subtypes (i.e. acral lentiginous melanoma in Black patients), direct comparison between demographic groups may not be a perfect comparison. Taking these limitations into consideration may at least partly explain the differences between the presented conclusions in this study and the findings of other studies related to the relationship between race and mortality and between sLDH and mortality [53,54].

Conclusions
This review of 334 stage IV melanoma patients revealed that elevated levels of sLDH, from as low as normal to 1.5 times ULN, are significantly associated with increased risk of mortality. Furthermore, Black patients and patients with lower incomes were found to have greater sLDH levels, so despite the insignificant difference in risk of mortality by race and income, further research is needed to assess the contribution of elevated sLDH levels in these populations to risk of mortality. Older age groups and presence of ulceration among patients with stage IV melanoma were also associated with a greater risk of mortality.
Author contributions AD and SC wrote the main manuscript text. CY and VS contributed to statistical analysis. RS, BP, and RS aided with idea curation and manuscript editing.
Funding None.

Declarations
Conflict of interest AD, SC, CY, VS, RS, BP, and RS all declare they have no conflicts of interest to report.
Ethical approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.