Liang Jianheng, Tan Qian, Wu Zheming, Chen Yun, Ju Ruihong, Chen Haisong, Tang Qiongyan, Xu Yanxue, Xu Li, Ortega-Usobiaga Julio, Cao Danmin, Wang Yong
J Refract Surg. 2025 May;41(5):e435-e443. doi: 10.3928/1081597X-20250328-01. Epub 2025 May 1.
To evaluate the accuracy of modern intraocular lens (IOL) power calculation formulas in eyes with previous myopic laser vision correction (LVC), and to assess the impact of axial length (AL) on the formula performance.
A total of 108 eyes were included, with 52 eyes in the AL less than 28 mm group and 56 eyes in the AL 28 mm or greater group. Refractive prediction errors (RPEs) were compared among nine post-LVC formulas: Haigis-TK, Haigis-TK (incorporating Cooke-modified axial length [CMAL]), PEARL-DGS, Hoffer-QST TK, LISA TK, Barrett True-K TK, EVO TK, Haigis-L, and Shammas. Pearson correlation analysis was employed to evaluate the influence of AL and other biometric parameters on RPEs.
In the AL less than 28 mm group, no significant differences were observed between all formulas (all adjusted > .05). However, in the AL 28 mm or greater group, Haigis-TK demonstrated significantly lower root mean square absolute error (RMSAE) and mean absolute error (MAE) compared to the Hoffer-QST, Barrett True-K TK, Shammas, and Haigis-L (all adjusted < .001). Similarly, the PEARL-DGS showed significantly lower RMSAE than the Barrett True-K TK, Shammas, and Haigis-L (both adjusted < .001), and significantly lower MAE compared to the Barrett True-K TK and Haigis-L (both adjusted < .001). The Haigis-TK had the highest percentage of eyes with RPEs within ±0.50 diopters (D) (73.21%), whereas the PEARL-DGS had the highest percentage of eyes with RPEs within ±1.00 D (94.64%). A significant negative correlation between AL and RPE was found in most formulas, leading to a myopic shift in eyes with extremely long AL.
The performance of current post-LVC formulas was comparable in eyes with AL less than 28 mm, whereas the Haigis-TK and PEARL-DGS demonstrated superior accuracy in eyes with AL 28 mm or greater. A notable myopic shift occurred in post-LVC eyes with extremely long AL, highlighting the need for careful formula selection in such cases. .
评估现代人工晶状体（IOL）屈光度计算公式在既往接受过近视激光视力矫正（LVC）的眼中的准确性，并评估眼轴长度（AL）对公式性能的影响。
共纳入108只眼，其中眼轴长度小于28mm组52只眼，眼轴长度28mm及以上组56只眼。比较了9种LVC后公式的屈光预测误差（RPE）：Haigis-TK、Haigis-TK（纳入库克修正眼轴长度[CMAL]）、PEARL-DGS、Hoffer-QST TK、LISA TK、Barrett True-K TK、EVO TK、Haigis-L和Shammas。采用Pearson相关分析评估AL和其他生物测量参数对RPE的影响。
在眼轴长度小于28mm组中，所有公式之间均未观察到显著差异（所有校正P>.05）。然而，在眼轴长度28mm及以上组中，与Hoffer-QST、Barrett True-K TK、Shammas和Haigis-L相比，Haigis-TK的均方根绝对误差（RMSAE）和平均绝对误差（MAE）显著更低（所有校正P<.001）。同样，PEARL-DGS的RMSAE显著低于Barrett True-K TK、Shammas和Haigis-L（两者校正P<.001），其MAE显著低于Barrett True-K TK和Haigis-L（两者校正P<.001）。Haigis-TK在RPE±0.50屈光度（D）范围内的眼比例最高（73.21%），而PEARL-DGS在RPE±1.00 D范围内的眼比例最高（94.64%）。在大多数公式中发现AL与RPE之间存在显著负相关，导致眼轴极长的眼出现近视偏移。
当前LVC后公式在眼轴长度小于28mm的眼中性能相当，而Haigis-TK和PEARL-DGS在眼轴长度28mm及以上的眼中显示出更高的准确性。在眼轴极长的LVC后眼中出现了明显的近视偏移，突出了在此类情况下仔细选择公式的必要性。
