Patients with WD may become symptomatic at any stage of life, with most reported symptoms occurring between the ages of 5 and 3516. According to our consecutive cohort of 1302 patients, approximately 92.70% (1207/1302) of Chinese patients with WD develop symptoms under the age of 30 years, 2.84% (37/1302) of the patients had symptomatic onset between 35 and 39 years old, and only 2.15% (28/1302) of the patients became symptomatic at ≥ 40 years old (Supplementary Table 6). The percentage of patients with an onset age ≥ 40 years old is significantly lower than those of European patients13,17. In Europe, a symptomatic onset age of 40 years has been set as the cut-off age for late-onset WD, resulting in 3.8% or 8% of patients being classified as having late-onset WD13,17. Considering that the majority of Chinese patients with symptomatic WD have an onset age under 30 years old and that there is a 4.5-year difference in onset age between patients with neurological manifestations in China (19.3 ± 8.0) and Europe (23.8 ± 9.1)3,6, we chose an onset age of 35 years as the cut-off age for late-onset WD. Based on these criteria and our large WD database (n = 2132), 105 patients (4.92%) were included, which facilitated our investigation of the clinical and genetic characteristics of late-onset WD.

In the present study, most patients with late-onset WD presented with neurological symptoms, and there was a greater proportion of hepatic presentation patients exhibiting liver cirrhosis, which is consistent with the findings of a recent study in France17. In East Asia, a predominance of male patients in the WD cohort has been observed in several studies6,18,19,20; however, we identified comparable numbers of male and female patients with late-onset WD. As less estrogen may be secreted in older patients, our data support the hypothesis that it plays a protective role21. Overall, 97 index patients were diagnosed based on K–F rings, serum ceruloplasmin levels, and extrapyramidal symptoms/abnormal brain MRI, and 2 of the patients were diagnosed based on additional 24 h urinary copper. Our data showed that 97.91% (94/96) of patients with late-onset WD had ceruloplasmin concentrations lower than 0.2 g/L, and 95.96% (95/99) of the patients had positive K–F rings, highlighting their role in diagnosing late-onset WD. However, the serum ceruloplasmin levels were significantly greater in the older onset patients (Table 2), and abnormal brain MRI could only be identified in 48.28% of patients with hepatic presentations; additional laboratory data, such as 24 h urinary copper, should help establish the diagnosis. In total, 91.92% of the index patients (91/99) were genetically diagnosed with two potentially pathogenic variants, and 7.07% of the patients (7/99) had one potentially pathogenic variant, with a total of 62 variants identified. As large hemizygous deletions, variants within the promoter, 3′ untranslated region, and far intronic regions of ATP7B have not been screened, not all variants may have been detected in the current cohort.

The top six most common variants were R778L, A874V, V1106I, P992L, R919G, and T935M in the late-onset cohort, which displayed significant differences in allele frequencies compared to those in patients with early-onset WD. Our data strongly suggest that A874V, V1106I, and T935M are associated with late-onset WD, with V1106I and T935M exhibiting low penetrance. According to gnomAD, the V1106I and T935M variants exhibit allele frequencies comparable to those of R778L in the East Asian subpopulation, and there is also a high allele frequency of R919G. However, there were no homozygotes of V1106I, T935M, or R919G identified in multiple large Chinese cohort studies6,19,22,23, and there were also no variant combinations of A874V/T935M, V1106I/T935M, V1106I/R919G, or T935M/R919G identified in the present study (Fig. 3b). The absence of these variant combinations in patients with WD implies that they may be nonpathogenic or that patients with these genotypes may be overlooked due to atypical clinical features. However, this conclusion was established based on several premises. One concern is that the genetic background of the East Asian subgroup in gnomAD may not represent the Chinese population, as there are differences in hotspot variants among different regions of China6. Other issues include a lack of Hardy–Weinberg equilibrium, the relatively small number of patients with late-onset WD included in the current study, and the limited number of patients with WD with acquired genetic information; therefore, these results should be confirmed in more patients with WD.

Our previous data support the conclusion that R778L, PTV, and P992L are associated with the early onset of WD, and the current data indicate that approximately 58.06% of these variants (36/62) occurred in combination with A874V, V1106I, T935M, or R919G in patients with late-onset WD. Additionally, we found that two patients carried a variant combination of R778L and PTV (R778L/L1395P-fs and R778L/V937G-fs), where L1395P-fs (c.4183dup, exon 21) was located at the end of ATP7B, and V937G-fs (c.2810del, exon 12) was reported to cause exon 12 skipping during ATP7B pre-mRNA processing24. Moreover, the alternative splice variant of ATP7B encodes a protein that lacks the entire TM5 domain and retains 80% of the biological activity of the wild-type protein, thereby eliminating the influence of the disease-causing variant in exon 1224, which may play an important role in the mild symptoms of WD. Variants R919G and T935M are also located in exon 12, and in vitro experiments have indicated that the T935M variant exhibits a trafficking pattern similar to that of wild-type ATP7B under copper pressure, which is distinct from that of the R778L, R919G, and P992L variants25. Notably, the R919G variant has also been shown to cause exon 12 skipping26, which may explain the higher allele frequency of the R919G variant in late-onset patients and raises speculation about the role of T935M and other exon 12 variants in the alternative splicing of ATP7B.

We noticed that the K866–H880 region, where the G869R, A874V, A874P, and G875R variants are located, is not present in monkey ATP7B (Supplementary Fig. 4), implying that this region may not be important for ATP7B function. This region is located within the flexible loop between the TM5 and A domains, and the A874V variant still exhibits partial copper transport and ATP hydrolysis7. The destabilization and endoplasmic reticulum (ER) retention caused by ATP7B variants also play key roles in the pathogenesis of WD7. However, the A874V variant displays different characteristics from other ER retention variants, since inhibition of the p38/JNK pathway involved in the degradation of the H1069Q and R778L variants, and copper treatment in correcting G875R localization both failed to rescue the A874V variant9,27. Another intriguing region is the N-domain of ATP7B, as it displays little sequence homology to other P-type ATPase families but appears to share the same core nucleotide-binding structure28. There are four invariant residues (E1064, H1069, G1099, G1101) in P1B-type ATPases, and the V1106I variant is located near segment P1098-G1103, which is mostly affected by ATP binding28. Our data indicated that V1106I is negatively associated with the neurological presentation of WD, implying that patients carrying the V1106I variant may exhibit a low tendency to accumulate copper in the brain. The low penetration of the V1106I variant in patients with WD indicates weak damage to the ATP7B protein, and it is possible that a large number of patients carrying the V1106I variant may exhibit no or only mild WD symptoms.

In conclusion, the current study highlighted the value of K–F rings, serum ceruloplasmin levels, and extrapyramidal symptoms in the diagnosis of patients with late-onset WD and highlighted the correlation between variants A874V, V1106I, T935M, and R919G and the late onset of WD. However, we must realize that serum ceruloplasmin levels in patients with late-onset WD are significantly higher than those in patients with early-onset WD, and a large number of patients carrying homozygotes of V1106I, T935M, and R919G or compound heterozygotes of A874V/T935M, V1106I/T935M, V1106I/R919G, or T935M/R919G may be missed because of atypical or a lack of WD symptoms. Additionally, in the late-onset cohort, 27.42% (17/62) of the variants were located in the K866–H880 region, TM5 domain, and N-domain of ATP7B, and 71.72% of the patients (71/99) carried at least one variant from these regions, indicating that these may be hotspot areas involved in the late presentation of WD. Our research revealed distinct clinical and genetic characteristics of patients with late-onset WD, which may provide valuable insights into the genetic basis and diagnosis of this disease.